200302761 (1) 玖、發明說明 【發明所屬之技術領域】 本發明是關於一邊協調動作廣範圍地移動工件上的加 工實行位置爲可能的低速定位手段與，具有X方向的邊長 DX以及Y方向的邊長DY的預定尺寸的加工區域內的窄 範圍地移動爲可能的高速定位手段，一邊加工散佈於工件 上的複數個加工位置時的加工計劃方法以及裝置，進行藉 由前述加工計劃方法決定的加工的加工方法、包含前述加 工計劃裝置的加工裝置、實施前述加工計劃方法或者實現 前述加工計劃裝置用的電腦程式以及記錄有該電腦程式的 可電腦讀取的記錄媒體。 【先前技術】 近年來伴隨著電子機器的小型化或高密度安裝化的要 求，重疊複數個印刷配線基板的多層印刷配線基板逐漸被 提供。在這種多層印刷配線基板，爲了電氣連接形成於疊 層於上下的印刷配線基板的每一個的導電層間，在這些基 板形成有稱爲貫通孔（through hole )或介層孔（via hole )的孔。而且，藉由在這些孔的內部形成導電膜以進行各 印刷配線基板的導電層間的連接。 形成於印刷配線基板的孔伴隨著最近的印刷配線基板 的小型化或高功能化而小型化，成爲直徑0.1mm以下。爲 了精度佳地形成這種小徑的孔，使用脈衝振盪型的雷射束 (2) (2)200302761 使用習知的脈衝振盪型雷射的雷射開孔機的一般的構 成如圖1所示。本構成例具備：包含由未圖示的雷射振盪 器照射的例如用以在預定的方向（在圖1垂直於紙面方向 )掃描脈衝狀的雷射光線20的旋轉鏡23的第一電掃描器 (galvano scanner) 22，與包含藉由該第一電掃描器22用 以在與藉由該第一電掃描器22的掃描方向垂直的方向（ 在圖1爲與紙面平行的方向）掃描在垂直於紙面的方向掃 描的雷射光線的旋轉鏡25的第二電掃描器24，與藉由該 第一以及第二電掃描器22、24將在兩方向掃描的雷射光 線固定於XY台12上的對基板等的加工對象物（稱爲工 件（work )) 10的表面垂直的方向偏向照射用的f- 0透 鏡2 6。 如此，藉由使用該第一、第二電掃描器22、24使雷 射光線20反射到掃描器前端的旋轉鏡23、25，可任意地 改變進行方向。此處，因旋轉鏡23、25爲輕量，故高速 定位爲可能。 藉由前述電掃描器22、24偏向的雷射光線通過f- 0 透鏡26，集光於工件10。此f- 0透鏡26因一般爲高價， 故尺寸被限定，因此，某時序（Timing)中的束（beam) 照射範圍（稱爲加工區域（area ))的大小被限制於數十 mm角左右的長方形，比一般的工件1 〇的大小還窄。 因此，藉由利用X Y台1 2傳送工件1 〇使廣範圍的定 位爲可能。但因XY台12的重量大，故定位所花的時間 大0 -6 - (3) (3)200302761 如此，雷射開孔機藉由使用高速窄範圍的定位裝置的 電掃描器以及低速廣範圍的定位裝置的XY台的兩個定位 裝置，進行高速廣範圍的開孔定位。 因此，由定位的觀點可大大地將動作形態（控制形態 )分類爲以下的兩個。 (1) 、電掃描器、XY台的各單獨控制 即透過串聯地進行藉由在使X Y台停止的狀態下的加 工區域內的雷射光掃描的雷射加工，與藉由χγ台的工件 的傳送，進行基板全體的加工的形態，通常被稱爲步進且 重複（step and repeat )。以下也稱爲低速定位手段停止 加工或台停止加工。 (2) 、電掃描器與XY台的協調控制（也稱爲同步控制） 並聯地進行藉由電掃描器的雷射光的掃描與藉由χγ 台的工件的傳送的形態，通常稱爲協調控制或同步控制。 以下也稱爲低速定位手段非停止加工或台非停止加工。 前述步進且重複揭示於例如美國專利第6087625號， 協調控制揭示於日本特開平3-266006等。 但是，美國專利第6087625號是關於步進且重複，並 非本案包含作爲對象的協調控制中的加工計劃。 而且’日本特開平3 — 266006是協調控制粗動台與微 動台’與本案作爲對象者完全不同。 (4) (4)200302761 【發明內容】 本發明其課題是爲了應解除前述習知的問題點所進行 的創作，在開始加工的前階段中，有效地計劃窄範圍地移 動工件上的雷射照射位置爲可能的高速定位手段，與廣範 圍地移動爲可能的低速定位手段的動作。 本發明是藉由一邊協調動作廣範圍地移動工件上的加 工實行位置爲可能的低速定位手段與，具有X方向的邊長 DX以及Y方向的邊長DY的預定尺寸的加工區域內的窄 範圍地移動爲可能的高速定位手段，一邊加工散佈於工件 上的複數個加工位置時的加工計劃方法，其中包含：藉由 如X或Y的加工時低速定位手段停止方向Z的邊長爲DZ 以下的矩形，分割加工位置的加工矩形分割製程；針對由 加工矩形分割製程決定的各加工矩形，使矩形內的總加工 時間最小，決定加工矩形內的加工位置訪問順序以及低速 定位手段動作速度的加工位置訪問順序以及低速定位手段 動作速度最佳化製程；以及藉由該兩個製程決定的以加工 矩形的位置、加工矩形入出口的低速定位手段動作速度之 中至少一個爲基礎，決定低速定位手段的動作路徑的低速 定位手段動作路徑最佳化製程，以解決前述課題。 而且，令該加工矩形分割製程（手段）是藉由X方向 或Y方向的單純的分割所進行的。 而且，令在進行該X方向或γ方向的單純的分割後 ，用以去除加工位置不存在的加工矩形，限定低速定位手 段的動作位置於具有加工位置之處’高速化加工以可實現 -8 - (5) (5)200302761 有效的協調控制。 而且，令該分割的加工矩形之中，在低速定位手段停 止方向Z的邊長之中若有比該方向的預定尺寸的邊長還小 者，不改變分割數而調整停止方向的邊長，消除停止方向 邊長小的加工矩形。 而且，令該加工矩形分割製程（手段）是藉由X方向 或Y方向的最少數的分割所進行的。 而且，令該加工矩形分割製程（手段）是在兩方試行 X方向的分割與Y方向的分割後，選取佳的（例如低速定 位手段的動作距離短者），高速化加工以可實現有效的協 調控制。 而且，令在進行該加工矩形分割後在各加工矩形中， 若沿著低速定位手段的動作方向加工位置具有無低速定位 手段的動作方向的預定尺寸的加工區域的邊長與速度變化 所需的距離和以上的位置的話，二等分該矩形。 而且，令在進行該加工矩形分割後在各加工矩形中， 令加工位置對與低速定位手段的動作方向垂直的方向集合 於中央而微調整加工矩形的配置。 而且，令該加工矩形分割製程（手段）在兩方試行X 方向的分割與Y方向的分割確定加工矩形後，藉由這些的 組合（例如低速定位手段的動作距離短者）加工矩形分割 全體，高速化加工以可實現有效的協調控制。 而且，令該加工矩形分割製程（手段）是用以考慮加 工位置的分布圖案（pattern)以及密度，藉由考慮加工位 (6) (6)200302761 置的疏密的矩形分割，疏的位置在疏的位置，密的位置在 密的位置特定矩形，藉由在疏的位置以快的速度，在密的 位置以慢的速度進行加工，高速化加工以可實現有效的協 調控制。 而且，令考慮該加工位置的分布圖案以及密度的加工 矩形分割製程（手段）包含：識別加工位置爲密的位置， 確定加工矩形的配置的製程（手段）；識別加工位置爲大 多數的位置，確定加工矩形的配置的製程（手段）；以及 對加工矩形未確定的加工位置，確定加工矩形的製程（手 段）。 而且，令該確定密的位置以及大多數位置的加工矩形 的配置的製程（手段）包含：分割工件上的加工區域成格 子狀，識別各格子內的該加工位置的疏密或有無的真僞的 製程（手段）：以所識別的格子的配置爲基礎，形成複數 個格子群矩形的製程（手段）：以及用以被覆各格子群矩 形而配置加工矩形的製程（手段）。 而且，令該格子群矩形在相當於矩形的外周的格子列 有被識別爲必定真的格子，在相當於矩形的外周外側的格 子列的格子均爲被識別爲僞的格子。 而且，令在識別加工位置爲大多數的位置，確定加工 矩形的配置的製程（手段）中，以對資料自身作爲圖案的 重複而記載的加工位置的擴大當作大多數的位置而識別。 而且，令在該用以被覆各格子群矩形而配置加工矩形 時，不滿足設定的大小的格子群矩形除外，決定加工矩形 -10- (7) (7)200302761 配置。 而且，令在該用以被覆各格子群矩形而配置加工矩形 的製程（手段）中，識別格子群矩形的直線狀或格子狀的 配置，決定加工矩形配置。 而且，令最終剩餘的加工位置的存在位置少的情形適 用使該低速定位手段停止，使高速定位手段動作的低速定 位手段停止加工，使加工有效化。 而且，令以與低速定位手段非停止加工相同的控制系 實行該低速定位手段停止加工，使設計容易化。 而且，令該加工位置訪問順序以及低速定位手段動作 速度最佳化製程（手段）是一邊最佳化藉由該高速定位手 段的移動、定位以及加工爲可能的加工位置的訪問順序， 一邊求出該低速定位手段的等速的速度的最大値的加工位 置訪問順序最佳化以及低速定位手段等速動作速度最大化 製程（手段）。 而且，令該加工位置訪問順序以及低速定位手段動作 速度最佳化製程（手段）在該加工位置訪問順序最佳化以 及低速定位手段等速動作速度最大化製程（手段）之後， 更對所決定的加工位置依照加工位置的分布密度，進行決 定平順地使動作速度變化的加減速模型的速度模型決定製 程。 而且，令在該加工位置訪問順序最佳化以及低速定位 手段等速動作速度最大化製程（手段）中，根據低速定位 手段動作速度，決定規定各加工位置的加工時序（Timing) -11 - (8) (8)200302761 於最早開始時刻以及最遲完了時刻之間的時間框。 而且，令在該加工位置訪問順序最佳化以及低速定位 手段等速動作速度最大化製程（手段）中，根據加工位置 間的移動量以及低速定位手段動作速度，決定加工位置間 的移動長。 而且，令該各加工位置的時間框以使用X或γ的預 定尺寸的加工區域的低速定位手段動作方向z的邊長DZ 、低速定位手段動作速度V、由該加工區域的加工開始時 刻TS中的加工位置P到預定尺寸的加工區域的最近邊的 距離PZ，最早開始時刻爲TS + PZ/V，最遲完了時刻爲TS + (PZ + DZ ) /V 決定。 而且，令該各加工位置的時間框以使用X或Y的預 定尺寸的加工區域的低速定位手段動作方向Z的邊長DZ 、低速定位手段動作速度V、由該加工區域的加工開始時 刻TS中的加工位置P到預定尺寸的加工區域的最近邊的 距離PZ，比DZ還小的預定的正或0的値DZF、DZB，最 早開始時刻爲TS+ ( PZ + DZF) /V，最遲完了時刻爲TS+( PZ + DZ-DZB ) /V 決定。 而且，令在該加工位置訪問順序最佳化以及低速定位 手段等速動作速度最大化製程（手段）中，在一個之前的 加工位置中的加工完了後，若加工位置間的移動完了的時 刻比最早開始時刻還早，在最早開始時刻若爲最早開始時 刻以後的話，決定各加工位置的加工開始時刻爲移動完了 的時刻。 -12- (9) 200302761 而且，令該加工位置訪問順序最佳化以及低速 段等速動作速度最大化製程（手段）是由使加工位 順序發生的加工位置訪問順序發生製程（手段）； 定加工位置訪問順序算出低速定位手段的等速動作 低速定位手段動作速度算出製程（手段），構成。 而且，令該加工位置訪問順序最佳化以及低速 段等速動作速度最大化製程（手段）是全包含使加 訪問順序發生的加工位置訪問順序發生製程（手段 定加工位置訪問順序算出低速定位手段的等速動作 低速定位手段動作速度算出製程（手段）；以及固 定位手段動作速度，改善加工位置訪問順序的加工 問順序改善製程（手段）。 而且，令該加工位置訪問順序最佳化以及低速 段等速動作速度最大化製程（手段）是由使加工位 順序發生的加工位置訪問順序發生製程（手段）； 工位置訪問順序算出低速定位手段的等速動作速度 定位手段動作速度算出製程（手段）：以及重複一 數次固定低速定位手段動作速度，改善加工位置訪 的加工位置訪問順序改善製程（手段）與該低速定 動作速度算出製程（手段）的重複改善製程（手段 成。 而且，令該加工位置訪問順序最佳化以及低速 段等速動作速度最大化製程（手段）是由使加工位 順序發生的加工位置訪問順序發生製程（手段）； 定位手 置訪問 以及固 速度的 定位手 工位置 );固 速度的 定低速 位置訪 定位手 置訪問 固定加 的低速 次至複 問順序 位手段 )，構 定位手 置訪問 不伴隨 -13- (10) (10)200302761 著加工實行可能性的判定，重複實行加工位置訪問順序改 善製程（手段）的無判定的加工位置訪問順序改善製程（ 手段）；以及 固定加工位置訪問順序算出低速定位手段的等速動作 速度的低速定位手段動作速度算出製程（手段），構成。 而且，令由該加工位置訪問順序發生製程（手段）構 成的加工順序是對該低速定位手段的動作方向設定寬度進 行加工矩形的分割，在該寬度內將容納於該寬度的加工位 置分類於與動作方向垂直的方向，若在某寬度內爲升順分 類的話，藉由使在其相鄰的寬度內爲降順分類，使加工位 置的訪問路徑成爲蛇行狀的順序。 而且，令根據加工矩形內的加工位置的密度，設定對 該低速定位手段動作方向設定寬度分割成帶狀的矩形時的 分割寬度。 而且，令該重複改善製程（手段）爲針對加工實行不 可能的高速側低速定位手段動作速度以及可能的低速側動 作速度之間的値V，首先判定是否可實行，可實行的情形 以V更新低速側動作速度，不可實行的情形針對V重複 實行加工位置訪問順序改善製程（手段）， 成爲可實行的情形以V更新低速側動作速度，到最後 不成爲可實行的情形，到重複次數比設定次數還多爲止或 者高速側動作速度與低速側動作速度的差比設定値還小爲 止重複以V更新高速側動作速度的處理之製程（手段）。 而且，令該重複改善製程（手段）爲針對加工實行可 -14- (11) (11)200302761 能的低速定位手段動作速度V，重複實行加工位置訪問順 序改善製程（手段），到重複次數比設定次數還多爲止， 或在加工位置訪問順序改善製程（手段）的前後總加工時 間無法縮短爲止，或在加工位置訪問順序改善製程（手段 )的前後低速定位手段動作速度的改善量比設定値還小爲 止，重複在改善解消失的時點更新可加工實行的動作速度 的處理之製程（手段）。 而且，令該加工位置訪問順序改善製程（手段）適用 附時間框非對稱哈密頓路徑長最小化問題或附時間框對稱 哈密頓路徑長最小化問題的解法。 而且，令該附時間框對稱或非對稱哈密頓路徑長最小 化問題的解法使用〇r-Opt法。 而且，令該加工位置訪問順序改善製程（手段）以及 該低速定位手段動作速度算出製程（手段）中的加工實行 可能性是根據所有的加工位置的加工完了時刻以及最遲完 了時刻的差來判定。 而且，令該加工位置訪問順序改善製程（手段）以及 該低速定位手段動作速度算出製程（手段）中的加工實行 可能性爲若所有的加工位置的加工完了時刻比加工終了的 最遲時序（Timing)的最遲完了時刻還早的話判定爲實行 可能，否則判定爲實行不可能。 而且，令該加工位置訪問順序改善製程（手段）中的 改善解的判定是根據在加工矩形內最後訪問的加工位置的 加工完了時刻以及該加工實行可能性的至少一方。 -15- (12) 200302761 而且，令該加工位置訪問順序以及低速定位手段 速度最佳化製程（手段）爲首先決定矩形內的加工位 問順序，其次，最佳化所決定的路徑中的該低速定位 的等速動作。 而且，令該速度模型決定製程（手段）是高速化 矩形內的加工位置的分布爲疏的部分中的該低速定位 的動作速度。 而且，令該加工位置訪問順序以及低速定位手段 速度最佳化製程（手段）的結果，若所有的加工位置 加工完了時刻比該最遲完了時刻還早的話，在所有的 位置中於加工完了時刻比最遲完了時刻還早的範圍朝 同一時間份的時刻方向偏移所有的加工位置的加工開 刻。 而且，令該低速定位手段動作路徑最佳化製程（ )是使移動加工矩形間的時間的總和最小化。 而且，令該低速定位手段動作路徑最佳化製程（ )是解決都市爲對應各加工矩形的入口與出口的位置 〜個，在相當於矩形的入出口的都市間的枝具有除去 的限制的巡迴售貨員問題或哈密頓路徑長最小化問題 程。 而且，令該低速定位手段動作路徑最佳化製程（ )是在最佳化加工矩形的訪問順序後，決定各矩形的 Q。 而且，令該低速定位手段動作路徑最佳化製程（ 動作 置訪 手段 加工 手段 動作 中的 加工 ，晚 始時 手段 手段 的每 禁止 的製 手段 入出 手段 -16- (13) (13)200302761 )是在決定各矩形的入出口後，最佳化矩形的訪問順序。 本發明是一邊協調動作廣範圍地移動工件上的加工實 行位置爲可能的低速定位手段與，具有X方向的邊長D X 以及Y方向的邊長DY的預定尺寸的加工區域內的窄範圍 地移動爲可能的局速定位手段，一邊加工散佈於工件上的 複數個加工位置時的加工計劃方法，包含：最佳化工件全 體的加工位置的訪問順序的全加工位置訪問順序最佳化製 程（手段）；以及對由該全加工位置訪問順序最佳化製程 (手段）決定的加工位置訪問順序，決定低速定位手段的 動作模型的低速定位手段動作模型決定製程（手段）。 而且，令該低速定位手段動作模型決定製程（手段） 首先以與預定尺寸矩形同尺寸的矩形群區分加工位置，接 著決定對應各群的加工位置的低速定位手段控制目標位置 〇 而且，一邊協調動作廣範圍地移動工件上的加工實行 位置爲可能的低速定位手段與，具有X方向的邊長DX以 及Y方向的邊長DY的預定尺寸的加工區域內的窄範圍地 移動爲可能的高速定位手段，一邊加工散佈於工件上的複 數個加工位置時的加工計劃方法，包含：藉由如X或Y的 加工時低速定位手段停止方向z的邊長爲DZ以下的矩形 ，分割加工位置的加工矩形分割製程（手段）；針對由該 加工矩形分割製程決定的各加工矩形’使加工位置的訪問 順序最佳化的加工位置訪問順序最佳化製程（手段）；以 及對由該加工位置訪問順序最佳化製程（手段）決定的加 -17- (14) 200302761 工位置訪問順序’決定該低速定位手段的動作模型的低速 定位手段動作模型決定製程（手段）。 本發明是提供一種加工方法’其特徵爲進行由前述任 一項所述之加工計劃方法決定的加工。 而且，提供一種用以實施則述任一項所述之加工計劃 方法的電腦程式。 而且，提供一種包含前述任一項所述之加工計劃裝置 的加工裝置。 而且，提供一種用以實現前述任一項所述之加工計劃 裝置的電腦程式。 而且，提供一種記錄有前述電腦程式的可電腦讀取的 記錄媒體。 【實施方式】 以下參照圖面詳細地說明適用藉由具備XY台（低速 定位手段）與電掃描器（高速定位手段）的雷射開孔機開 孔的情形的本發明的實施形態。 首先，簡單地說明成爲發明的基礎之構想。 在發明者已經提出的日本特開平200 1 -33 1 550中提出 爲了高速實現藉由步進且重複的動作（即低速定位手段（ 台）停止加工），用以在加工前計劃χγ台（也有僅稱爲 台）的停止位置、停止位置的訪問順序以及電掃描器的掃 描軌道等的技術。 此爲關於以下的三項目進行最佳化者。 -18- (15) (15)200302761 < 1 >、若爲開孔位置（加工位置）的加工區域（若爲 被照射雷射束的範圍則形狀不限定於矩形）分割（XY台 動作次數的降低） <2>、加工區域內的開孔位置訪問順序（掃描器動作 時間的降低） <3>、加工區域的訪問順序（XY台動作量的降低） 本發明是使進行低速定位手段停止加工（此處爲台停 止加工）用的上述構想擴張到低速定位手段非停止加工（ 此處爲台非停止加工），關於以下的四項目用以最佳化者 〇 (1 )、開孔位置的加工矩形分割（XY台動作位置的 限定） (2 )、加工矩形內開孔位置訪問順序（掃描器動作 時間的降低） (3 )、加工矩形通過時的XY台速度（也包含加減 速模型）（XY台的等待時間降低） (4 )、加工矩形訪問順序以及加工矩形的出入口（ XY台動作量的降低） 即爲了台非停止加工，台停止加工的<2>的最佳化成 爲分成關於（2 )與（3 )的兩項目的最佳化的形。 此處，[加工矩形]是指成爲XY台爲非停止一邊在一 方向直進’一邊加工時的加工的單位的區域，加工區域沿 著XY台動作方向延伸的形狀（長方形）。 此外’此處雖然低速定位手段以XY台，高速定位手 -19- (16) (16)200302761 段以使用電掃描器的情形爲例來說明，惟低速定位手段或 高速定位手段的種類並非限定於此。 關於台非停止加工的（2 )的加工矩形內開孔位置訪 問順序可考慮由於（3 )的XY台速度的大小使最佳的順 序不同。相反地，（3 )的XY台速度的最大速度其値會 因（2 )的加工矩形內開孔位置訪問順序的好壞而不同。 因此，因上述四項目之中（2 )與（3 )密切地相關，故一 起最佳化而不是獨立地最佳化較佳。 因此，如圖2所示分成使全體獨立的三個製程。 A、 開孔位置的加工矩形分割製程（步驟1 00 ) B、 （各加工矩形內的）開孔位置訪問順序以及XY 台動作速度最佳化製程（步驟110) C、 XY台動作路徑最佳化製程（步驟120) 步驟1 00的開孔位置的加工矩形分割最佳化的槪念顯 示於圖3。本製程是藉由在與XY台進行方向（實際上在 基板上的加工區域遷移方向，台在相反方向動作，惟如圖 爲了使寫箭頭者容易暸解起見，寫成台進行方向）垂直的 方向爲加工區域的該方向的邊長以下的長度的與基板的配 置方向平行的矩形（加工矩形），分割開孔位置的製程。 步驟11 0的各加工矩形內的開孔位置訪問順序以及 XY台動作速度最佳化製程的槪念顯示於圖4。本製程是 關於由步驟1 00決定的各加工矩形，使加工矩形的加工時 間即最後的加工位置的加工（雷射照射）完了時刻與最初 的加工位置的加工開始時刻的差最小，一邊最大化XY台 -20- (17) (17)200302761 的動作速度（不爲一定速度也可以），一邊最佳化加工矩 形內的開孔位置訪問順序的製程。 βρ X Y台以等速動作的情形加工區域的大小的問題存 在’對各開孔位置即使開始雷射照射也可以的時序（ Timing)，與照射必定必須完了的時序（Timing)(稱爲時 間框限制）。例如若χγ台的速度極慢的話，雖然不引起 所有的開孔位置的照射來不及的狀況，但加工時間變長。 另一方面，若XY台的速度過快的話，會發生電掃描器的 掃描造成的定位來不及的狀況。 而且，掃描器的掃描由於各掃描的掃描距離，因開孔 位置的訪問順序的好壞使花費時間被左右。 由以上，需求出利用掃描器的掃描、定位以及照射爲 可能（以下稱對加工矩形內的所有的開孔位置可實行的照 射時序（Timing)被決定爲實行可能）的開孔位置的訪問 順序以及XY台最大速度。 在此製程中開孔位置的訪問順序、各開孔位置的照射 時序（Timing)、XY台速度、加減速度模型等有決定必要 的要素有無數個，爲極爲困難的製程，故分成如以下的兩 個製程。 B 1、以等速使XY台動作的情形的決定像動作速度最 大的加工矩形內的開孔位置訪問順序的製程（步驟1 1 2 ) 。以下記爲開孔位置（加工位置）訪問順序最佳化以及 XY台（低速定位手段）等速動作速度最大化製程。 B2、對上述被決定的開孔位置，由開孔位置的分布密 -21 - (18) (18)200302761 度決定使動作速度平順地變化的XY台加減速模型的製程 (步驟1 1 4 )。 此外，由加工精度、基板的開孔位置的圖案、使用的 控制系的設計形態等可判斷台以等速使台動作者較佳的情 形，步驟1 14跳過。 步驟1 20的ΧΥ台動作路徑最佳化製程的槪念顯示於 圖5。本製程是以由步驟1 〇〇決定的加工矩形的位置以及 由步驟110決定的加工矩形入出口的ΧΥ台動作速度爲基 礎，最佳化ΧΥ台的動作路徑（加工區域的加工矩形訪問 順序以及各加工矩形上的加工區域遷移方向）的製程。 以下，針對各製程詳細地說明。 前述步驟1 00的開孔位置的加工矩形分割最佳化製程 是藉由與χγ台進行方向垂直的方向爲加工區域寬度以下 的長度的藉由與基板的配置方向平行的加工矩形，分割開 孔位置的製程。 實施此製程的方法考慮如以下的五個方法。 1、 藉由橫方向或縱方向的單純的分割的方法 2、 橫或縱的分割，去除孔不存在的位置的方法 3、 關於上述1、2，在兩方試行橫方向、縱方向後， 選取佳的方法 4、 關於上述1、2，兩方試行橫方向、縱方向確定加 工矩形後，藉由這些加工矩形的組合以加工矩形分割全體 的方法 5、 考慮同一圖案以及密度的更有效的加工矩形分割 -22- (19) (19)200302761 方法 以下，針對各個方法具體地說明。 1、藉由橫方向或縱方向的單純的分割的方法 如圖6，在X方向（橫方向）或γ方向（縱方向）預 先決定XY台的進行方向（圖爲橫方向），與台動作方向 (X方向）垂直的方向（Y方向）的加工矩形的一邊的大 小取加工區域的大小以下的値，台進行方向（X方向）的 一邊的大小取由開孔位置的最大値減去最小値的値，由基 板的端依序無間距地配置加工矩形的方法。 雖然未必爲最佳的分割，但實際上如圖7所示，因基 板的開孔位置常普遍擴大到全體，有即使嘗試最佳化也得 不到解，故本方法也爲有效的手法。 依照狀況可加入如以下的工夫。 a、 對開孔位置不存在的加工矩形的應付方法 橫方向或縱方向的單純的分割的情形，有完全無開孔 位置的加工矩形產生的情形。由於這種加工矩形不需要， 故進行去除的處理。 b、 與XY台進行方向垂直的方向的一邊的大小的設定方 法 假想XY台在橫（X )方向移動的情形。即令XY台 進行方向爲橫（X )，與XY台進行方向垂直的方向（停 -23- (20) (20)200302761 止方向）爲縱（Y )。令加工區域的大小爲加工時進行方 向爲DM，加工時停止方向爲DS ’基板的縱（Υ)寬（ 開孔位置的γ座標的最大、最小的差）爲τ。在此手法中 加工矩形的片數變成[(T_a ) /DS] + 1°此處’ 是表示 不超過該數Α的最大整數的高斯（Gauss )記號’ α指與 Τ、DS比較十分小，接近零的正値。（^爲T/DS剛好整 除的情形的應付法）。例如T = 400mm、DS = 50mm的情形 ，加工矩形的片數爲1〇片，對於T = 4〇〇mm、DS = 40mm的 情形，α = 0 · 0 0 1 m m還是1 0片。 此處，加工矩形的停止方向的長度與加工區域的大小 D S無須相同，加工矩形的一邊與加工區域的大小相同或 者爲小的値的話佳。因此，令基板的縱長度T除以加工矩 形的片數[(Τ- a ) /DS] + 1的値爲各加工矩形的縱長度。 藉由此手法，例如在T = 460mm、D = 50mm的基板中藉由每 460/1 0 = 46mm劃分，使像對於每50mm劃分的情形最後 l〇mm的寬度的加工矩形剩餘的現象消失。爲了實現機械 的限制或高精度的加工，由於通常XY台動作速度的界限 存在，故在寬度窄的加工矩形的加工中想避免顯著地XY 台動作速度接近界限速度的狀態，但藉由本手法因加工矩 形的寬度被均勻化，故可避免此點。 c、無開孔位置的位置的刪除與加工矩形分割 在圖8 ( A)所示的例子，在加工矩形內例如由左朝 右依次邊進行邊找開孔位置的話，完全無開孔位置的位置 -24- (21) (21)200302761 存在。這種位置即使以非停止使台動作因浪費，故如圖8 (B)所示想分割加工矩形。分割加工矩形的基準例如可 如以下所示。 甲、在加工矩形的左端或右端無點的情形 乙、當加工矩形朝右方向依次邊進行邊找開孔位置時 ，無[加工區域的寬度DM +速度變更所需的距離α ]以上的 點存在。 此處，令分割的基準爲無DM+ α以上的點的位置的理 由如以下。即如圖9 ( A )所示開孔位置的塊被二等分的 情形，調查加工矩形是否可分割成左右兩個加工矩形。 關於左邊的加工矩形的加工終了位置與右邊的加工矩 形的加工開始位置，左邊的加工矩形的加工終了位置需位 於比右邊的加工矩形的加工開始位置還左，惟各個位置如 圖9 ( B )所示由於是在書寫接鄰於加工矩形的左右的加 工區域時的中心位置，故即使最低也需排列於進行方向的 加工矩形間隔爲DM以上。 再者，由左邊的加工矩形的加工終了位置移動到右邊 的加工矩形的加工開始位置時，如圖9 ( C )所示需變更 加工矩形內的動作速度。因此，需要一點點動作距離α ( α的値是藉由所使用的台的特徵適切地決定）。 除此之外，加工矩形間的間隔最低需要DM+ α以上。 d、加工矩形位置的微調整方法 垂直於加工矩形的進行方向的方向（此處令Y方向） -25- (22) (22)200302761 的配置是使開孔位置集合於加工矩形的中央而可上下錯開 。即在某加工矩形對於使開孔位置集合P={Pi、pi + 1...pj ( i $ j )存在而矩形分割的情形，藉由令屬於P的點的Y座 標値的擴大的中心（最大値、最小値的平均）爲加工矩形 的縱方向的中央位置，可集中Υ方向的電掃描器的動作於 中央。或者也能以重心（重心偏上或下的情形，所有的點 爲不超出加工矩形的界限位置）而非擴大的中心爲中心。 2、橫或縱的分割，去除孔不存在的位置的方法 如圖1 0所示的去除無孔的位置進行加工矩形分割。 具體上，若假想ΧΥ台在橫方向移動的情形，則以此方法 加工矩形的位置不爲成爲如圖1 1 ( A)的段不同的配置， 而是如圖11 ( B)用以成爲整齊地一致的配置。因此，僅 使用開孔位置的Y座標資訊決定加工矩形的配置。 具體上定義如以下（X)的問題（稱爲SPLIT)，如 以下的（Y )使用SPLIT的解法處理問題。 問題（X ) (SPLIT ): 將被升順分類的η個數値資料的集合A = { a 1、a 2…a η} ，分割（分割是指對i古j的任意的i、j，Bi Π Bj = 0 (空 集合）成立，而且，m U wBi =A )成最少個數的互質的部 分集合（互質的兩個集合是指在集合的要素無同一者） Β!、B2…Bm。但是，Bk的要素被分類，Bk的末尾的要素必 定比B k +1的即頭的要素還小，各部分集合b k == { a s…a。}關於 前頭、末尾的要素滿足| ac - as | < = DS ( DS是表示掃描 -26- (23) (23)200302761 區域的大小的固定値）。 處理（Y ): <1>、僅取出基板全體的開孔位置的Y座標値（分割 縱軸的情形。分割橫軸的情形爲X座標値）不重複者而分 類，作爲集合A的要素（A={ai...an}，ai<ai + i)。 <2>、令 A爲問題SPLIT的輸入，得到部分集合 B 1 …B m。 <3>、以各部分集合B!... Bm爲基礎，將開孔位置分割 成m個群。 SPLIT的解法具體上問題爲以對圖的各邊附加權重時 的網路最佳化問題之一的求由給予的節點到所有的節點的 最短路徑的最短路徑問題來捕捉，適用公知的解法。全體 如圖1 2所示來處理。 即首先在步驟200如圖13 ( A )所示，僅取出開孔位 置的Y座標値不重複者而分類，作爲集合A={a:、a2... an} 的要素（此處，分類是指對1 g i S n-1的i，使ai < ai+1 而重排）。再者，如圖13 ( B )所示，設節點集合V={ v! 、v2…，令V的要素Vi其値附對應於ai。 而且，令兩節點（Vi、L) ( i<j )爲與下述的情形接 鄰者。 ai + D ^ aj 或者 ai + D g aj-i 且 ai + D<aj 即接鄰於Vi的Vj是指令對應Vi的値ai爲某加工矩形 -27- (24) (24)200302761 SQ1的下邊的Y座標時，使對應Vj的値^爲下邊的γ座 標的加工矩形可成爲與基板中的s Q 1相鄰的加工矩形。 再者’接鄰的兩節點（Vi、Vj ) ( i<j )對的權重是令 全部爲1 (權重是指兩點間的枝的長度（成本））。 而且’令始點爲Vi ( Y座標最小的點），終點候補集 合爲V e = { v k…v n}(但是k是指自k = 1依升順調查，成爲 ak +D ^ an的最初者，即令終點候補集合的要素爲下邊位 置的加工矩形包含在Y座標具有最大値an的點）。 由以上，在步驟2 0 2可得到令v 1爲始點的有向網路 〇 對所得到的網路，在步驟204使用例如像Dijkstra的 方法的已知的手法，解決最短路徑問題。於是，如圖丄3 (B )所示可求出由始點到各節點的最短路徑與最短路徑 長。 此外，Dijkstra的方法等爲的確可得到最短路徑（此 處爲最少的加工矩形）的手法。 接著，在步驟206選擇終點候補集合的各節點之中， 最短路徑長顯示最小値的節點。 其次，在步驟208若以對應朝該節點的最短路徑上的 節點的値爲下邊的Y座標値的加工矩形分割的話，如圖 14 ( C)所示，可得到想求的加工矩形分割。此處，最短 路徑長變成（加工矩形數-1 )。 而且，上述加工矩形縱位置的設定方法雖然以最少化 加工矩形的片數爲目的，但加工矩形的片數即使同一如圖 -28- (25) (25)200302761 1 5 ( A)所示，仍有藉由加工矩形被覆使開孔位置被分斷 於兩端（圖的上下）。因此，由最少的加工矩形片數的解 之中選擇開孔位置的分斷最無解者較佳。爲了實現此點， 對加工矩形片數爲最少的解全部，上述分斷的間距（gap )値比規定値還大的位置位於加工矩形內的情形’若加上 該間距値，選擇間距値的總和爲最小者佳。 或者另外的實現方法，加上加工矩形間的間距値即加 工矩形的最終要素與1段上的加工矩形的最初的要素間的 間距値，選擇間距値的總和爲最大者也可以。 此外，在本手法中也可加入對前述1、的C以及d的 各種各加工矩形的工夫。 關於藉由上述1、2的方法的加工矩形分割，圖2中 的步驟1 20的加工矩形動作路徑最佳化處理很簡單。即因 加工矩形成爲段一致的配置，故XY台進行方向爲橫向的 情形，若以由基板內的縱位置下方所具有的加工矩形依次 XY台的動作路徑（如已經說明的，正確爲加工區域遷移 路徑）由基板下方朝上方成爲蛇行狀般的路徑的話佳。此 情形的全體的處理如圖1 6所示。 即在步驟220中進行藉由1、或2、的方法的段一致 的矩形分割。 其次，在步驟222中令χγ台的動作路徑爲由基板下 方朝上方成爲蛇行狀而決定。 其次，在步驟224中進行各加工矩形的開孔位置的訪 問順序最佳化以及XY台動作速度最大化。 -29- (26) (26)200302761 此外，步驟222與224以相反的順序也無妨。 3、關於上述1、2，在兩方試行橫方向與縱方向後，選擇 佳者的方法 在此方法中如圖1 7所示，試行橫方向以及縱方向兩 方的加工矩形分割看看，求出各個動作距離，選擇動作距 離小者。在圖1 7因在橫方向分割的情形的加工矩形分割 其動作距離短，故選擇橫方向的加工矩形分割。 此處，雖然選擇動作距離小者，但選擇的基準例如考 慮開孔位置的疏密的方法也可以。即如圖1 8 ( C )所示， 在加工矩形內混合存在開孔位置疏的位置與密的位置的情 形，在密的位置需要降低XY台動作速度，譬如即使可在 同一加工矩形內使XY台動作速度可變，無論如何因受到 密的位置的速度的影響，故XY台動作速度模型決定變的 困難。 爲了避免這種狀況，在進行加工矩形分割的階段中開 孔位置疏的位置集中當作一個加工矩形，密的位置另外集 中當作一個加工矩形較佳。因此，具體上藉由例如如圖 1 9所示的方法，評價加工矩形的疏密的均勻度，選擇縱 橫佳者。 即首先在步驟300沿著XY台進行方向，作成累積頻 率圖。此處所請的累積頻率圖如圖2 0 ( C )所示，橫軸爲 距加工矩形的端的朝XY台進行方向距離，縱軸爲沿著 XY台進行方向調查開孔位置，表示對其加工矩形內的孔 30- (27) (27)200302761 數出現幾％的百分比的圖。 此累積頻率圖如圖20 ( C )所示爲單調遞增，而疏密 越一樣越成爲直線狀的單調遞增。因此，根據距離直線多 少，可評價疏密的一樣性。 因此，在步驟300對各加工矩形作成累積頻率圖，在 步驟302求出這些的平均的累積頻率圖。 接著，在步驟304以例如面積等評價距直線狀的增加 的偏移値。然後，在步驟306縱橫兩方求出該評價値，選 擇佳的。 此外，如上述以最大偏移的大小而非以面積來評價等 也能使用疏密的狀況成爲浮雕的其他手法。 或者針對X以及Y的兩方向的分割進行圖16的處理 ，根據在步驟224中計算的各加工矩形的加工時間，也能 選擇加工時間的總和小者。 4、關於上述1、2在兩方試行橫方向與縱方向確定加工矩 形後，藉由這些加工矩形的組合，加工矩形分割全體 的方法 在此方法中如圖21所示，試行橫方向優先與縱方向 優先兩方的加工矩形分割看看，在各個加工矩形確定後， 由這些之中使動作距離最小選擇兩方向的加工矩形的組合 〇 在圖2 1中僅左下的部分對縱方向的加工矩形分割因 動作距離短，故僅此部分在縱方向當作長的加工矩形，其 -31 - (28) (28)200302761 他部分在橫方向選擇長的加工矩形。 此外，不以評價的基準當作動作距離最小，如前述以 密度的一樣性來評價也可以。 而且，不以評價的基準當作動作距離最小，如前述以 各加工矩形的加工時間的總和來評價也可以。 5、考慮同一圖案以及分布密度的更有效的加工矩形分割 方法 在此方法，被分割的加工矩形內的開孔位置在一邊使 XY台動作一邊進行藉由電掃描器的定位的情形降低像不 利的情形，以得到更有效的加工矩形分割。 當分割矩形內的開孔位置的位置不爲有效的情形，如 圖1 5 ( A)所示有開孔位置二等分成上下的情形，與如圖 1 5 ( B)所示開孔位置的密度不一定，疏的位置與密的位 置混合存在的情形。 即如圖1 5 ( A )所示對於XY台進行方向爲橫向的情 形，開孔位置在加工矩形上邊近傍以及下邊近傍兩極化， 對於像在加工矩形中央近傍無的情形，因電掃描器的動作 量在縱方向多，故加工速度不提高。 而且，如圖15(B)所示開孔位置的分布密度不一定 ，對XY台進行方向疏、密、疏、密…變化激烈的情形， 疏的位置中的XY台的動作速度受到密的位置中的動作速 度的限制，很難決定較佳的速度模型。 因此，如上述的開孔位置的兩極化或密度不均勻狀態 -32- (29) (29)200302761 避免較佳。 而且如圖2 2可見到的，開孔位置的分布常呈同一或 類似的形狀（同一資料的反轉等）的開孔位置群（稱爲圖 案形狀）11 A爲複數個以及除此之外的若千剩餘點1 1 B的 分布的方式。 加工矩形分割困難的理由之一乃因上述若干數目的剩 餘點1 1 B存在，故這種情形並非在難解的狀態下直接進行 加工矩形分割，首先，區分圖案形狀1 1 A，其次關於其大 多數的圖案形狀1 1 A較容易地進行加工矩形分割，然後’ 對若干數目的開孔位置1 1 B進行加工矩形分割不僅可簡單 地應付，而且常成爲有效的加工矩形分割。 以下，參照圖23說明開孔位置的加工矩形分割的順 序。 首先，在步驟400中以GUI操作等依照預先指示的內 容，判定是否進行疏密的分別。 對於判定結果爲正的情形進行到步驟402，對於最疏 的群以外的開孔位置，可有效地確定加工矩形被確定。 其次，進行到步驟404，以GUI操作等依照預先指示 的內容，判定是否進行孔位置的大多數/少數分布的分別 〇 判定結果爲正的情形進行到步驟406，對於大多數群 的開孔位置，可有效地確定加工矩形被確定。 進行到步驟408，對加工矩形未確定的開孔位置確定 加工矩形。 -33- (30) (30)200302761 此外，步驟408使用已經敘述的1〜4的加工矩形分割 方法的任一個。 前述步驟402以及步驟406如圖24 (步驟402 )以及 圖25 (步驟406 )所示，具有類似的製程。即 (1 )、格子分割以及格子識別（位元分配）製程（ 步驟4 1 0 ) (2)、格子群矩形化製程（步驟414、415) (3 )、格子群矩形位置的微調整製程（步驟4 1 8、 4 1 9 )以及 (4)、藉由格子群矩形的加工矩形的被覆（加工矩 形的配置）製程（步驟420、421 ) 都以此順序而具有。 大致敘述各製程。步驟4 1 0、4 11的（1 )、格子分割 以及格子識別（位元分配）製程是將基板上的加工區域（ 即開孔位置的擴大區域）分割成格子狀，以〇或1 (位元 )識別各格子（在以下爲平行於大小一致的基板的矩形的 意思，僅稱爲格子）內的加工位置的疏密或有無的製程（ 稱識別値爲1的格子爲真的格子，識別値爲〇的格子爲僞 的格子）。 步驟414、415的（2)、格子群矩形化製程如圖26 所示，以如以下的條件的矩形（稱爲格子群矩形）群化分 配有1的格子的製程。 甲、在相當於矩形外周的格子列（上下左右四個）必 定存在1 (有點）。 -34- (31) (31)200302761 乙、包圍矩形外周的格子列（上下左右四個）均爲0 (無點）。 而且，步驟4 1 8、4 1 9的（3 )、格子群矩形位置的微 調整製程在（2 )、格子群矩形化製程因被在XY方向具 有寬度的格子區分，故例如孔的有無實際區分的點並不一 定，故爲更正確地調整格子群矩形的配置的製程。 步驟420、421的（4)、藉由格子群矩形的加工矩形 的被覆製程爲所得到的格子群矩形爲 甲、直線狀配置 乙 '格子狀配置 的情形，確定加工矩形的製程。 而且，在步驟4 1 6對於調查各格子群矩形的大小，若 無大的格子群矩形的情形，使步驟419以及421不進行。 此外，如圖25所示，步驟419以及421爲僅針對大的格 子群矩形進行的製程。 以下詳細地說明。 前述步驟4 1 0中的格子分割以及密格子的識別具體上 以如圖2 7所示的順序如圖2 8所示地進行。 即首先在步驟500分割基板成格子狀，計數各格子內 的孔數。其次，在步驟504作成孔數的度數分布，調查是 否有度數顯著地不同的位置。 此外’在格子狀的分割時，視原來的基板爲一個格子 ’若縱橫二等分該格子的話產生四個格子，惟藉由重複這 種分割可細分。 -35- (32) (32)200302761 即在步驟500每次反覆細分割，在步驟502計數格子 內孔數，在步驟504作成[孔數*以上*未滿的格子爲*個]的 要領的度數分布，在步驟5 06反覆該度數分布中的最大度 數爲到比某設定値還小。 其次，在步驟508對密格子分配1 ;對疏格子分配〇 〇 圖29是對格子分割基板，計數開孔位置的度數的情 形，僅密的位置11C或11D以影線網作出陰影的圖。此 外，在圖29以圓形包圍的密的位置1 1D判斷爲屬於在圖 25的步驟416中面積窄的格子群矩形的格子，將加工矩 形的決定放到圖23的步驟408的格子的例。 而且，在前述步驟4 1 1中的格子分割以及有孔格子的 識別具體上可藉由如圖30所示的順序如圖3 1所示地進行 〇 即首先在步驟606以適當大小的格子分割基板。其次 ，在步驟6 1 0如圖26所示即使是一個，也對有點的格子 分配1 ;對全無的格子分配0。 在前述步驟414、415中的格子群矩形化具體上以如 圖32所示的順序進行。 即首先在步驟622任意選擇格子群矩形未確定的分配 有1的格子，暫定地令該格子自身爲一個格子群矩形。 關於相當於所選擇的現在的格子群矩形外周（上下左 右）的外側的格子行或列（不包含相當於矩形的四角落的 格子）的每一個，在步驟624若判定有一個格子的話，在 -36- (33) (33)200302761 步驟626擴大格子群矩形的面積於其方向。在步驟624若 上下左右都沒有的話，在步驟628令格子群矩形確定，在 步驟630對於判斷具有分配有1的格子的情形返回到步驟 622，再度繼續從步驟622的處理，對於判斷無的情形使 處理終了。 前述步驟4 1 8、4 1 9中的格子群矩形位置的微調整具 體上分別以如圖3 3、圖3 4所示的順序進行。 即首先爲了調查這些矩形的實際的正確的開孔位置的 擴大，關於所有的矩形關於位於外周的格子列（上下左右 )，調查位於最外側的開孔位置。 即詳細顯示大圖案格子群矩形位置的微調整的步驟 419的圖34的情形，若求 左的格子列的開孔位置的X座標最小値（步驟644 ) 右的格子列的開孔位置的X座標最大値（步驟650 ) 下的格子列的開孔位置的Y座標最小値（步驟65 6 ) 上的格子列的開孔位置的Y座標最大値（步驟662 ) 的話，可求出實際的正確的開孔位置的擴大。若藉由 這些正確的最大、最小値更新顯示開孔位置的存在的矩形 的座標資訊的話，規則正確的配列的識別爲可能。雖已經 敘述，但對於原來的資料檔案自身像具有圖案資訊的情形 -37- (34) 200302761 ，無須進行如上述的處理。 此外，對於詳細地顯示密格子群矩形位置微調整的步 驟41 8的圖3 3的情形’有在判定爲疏分配有〇的位於格 子群矩形外周外側的格子列具有孔的可能性。 因此，在步驟640、646、6 52、658調查在位於格子 群矩形外周外側的格子列是否有孔’用以在有的情形下求 出外周外側格子列的最大値或最小値（步驟642、648、 654、660 )，在無的情形求出位於外周的格子列的最大値 或最小値（步驟644、650、656、662)。 而且，藉由在前述步驟420、421中的密或大的格子 群矩形的加工矩形的被覆（即加工矩形的配置），以如圖 35所示的順序進行。 此處，由格子群矩形的集合識別規則正確的排列，被 覆這些規則的矩形集合。 此處所謂的規則正確的排列是指如圖36 ( B )所示， 矩形配置成直線狀的關係或都在XY方向配置成直線狀的 格子的關係。因此，進行如以下的處理。 首先，在步驟670在X方向、Y方向的直線狀的配置 時，關於X方向、Y方向的各方向進行座標的分類以群化 値相同者。在圖36的例子X1{B5、BIO}、X2{B6、B11}、 X3{B7、B12}在 X 方向，Y1{B1、B2、B3、B4}、Y2{B5、 B6、B7}、Y3{B10、Bll、B12}在Y方向配置成直線狀。 其次，對於直線狀配置在X、Υ都有複數個的情形， 由這些的組合調查格子狀配置是否被製作。即在步驟676 -38- (35) (35)200302761 中關於上述被群化的矩形，調查如X方向的複數個群的要 素與Y方向的複數個群的要素一致的組合。在此例中X 方向的三群 X1{B5、 BIO}、 X2{B6、B11}、 X3{B7、 B12} 與 Y 方向的兩群 Y2{B5、B6、B7}、Y3{B10、B11、B12} 其要素一致。找到這種一致的組合用的實際的方法如以下 。即首先依次選擇X方向的群（最初爲X1{B5、B10})， 調查Y方向之中是否有具相同要素的群（對B5有Y2{B5 、B6、B7，對 B10 有 Y3{B10、B11、Β12Π。調查某情形 新產生的要素{B6、B7、Bll、B12}在X方向的XI以外的 群X2、X3是否具有的話佳。由以上的操作若完全一致的 話成爲格子狀的配置。 由以上直線狀配置爲丫1丨31、32、33、；64}，格子狀 配置爲 X Yl{ B5、B6、B7、BIO、Bll、B12}可求出。 對於有如此求出的格子狀配置的情形，在步驟680對 於有直線狀配置的情形，在比步驟680還後的步驟的步驟 674，用以被覆規則地配置的格子群矩形而配置加工矩形 〇 在被覆直線狀配置的矩形群時如圖36所示，注意以 下的點。 甲、基本上如圖37 ( A )所示，依照直線的延伸方向 決定加工矩形配置。 乙、如圖3 7 ( B )所示，格子群矩形的配置間隔（矩 形的最近位置之間的距離）若比以前述[1、藉由橫方向或 縱方向的單純的分割的方法]記述的（加工區域的大小 -39- (36) (36)200302761 DM+關於速度變更的距離α )還大的話，配置複數個加工 矩形。 丙、如圖3 7 ( C)所示，與直線方向垂直的方向的寬 度比加工區域的大小DS還廣，無法以一個加工矩形被覆 的情形，配置兩個以上的加工矩形。 丁、如圖3 7 ( D)所示，在配置複數個加工矩形的情 形若單純地配置的話，對於像開孔位置於端二等分的情形 ，順利地設定被覆加工矩形的中心位置。具體上，並非由 端依次以加工區域的寬度被覆，而是用以避開無開孔位置 的間距的部分，與應配置有開孔位置的部分的中心的加工 矩形的中心重疊。 另一方面，在被覆格子狀配置的矩形群時，如圖38 所示，注意以下的點。 與直線狀配置的情形一樣如圖3 8 ( A)所示，對於配 置間隔比D+ α還大的情形，在X方向、Y方向都分割矩 形群。其結果，排列成格子狀的格子群矩形其間隔視爲 D+α以下。如圖38(B)所示，以複數個X或Υ方向的 同一方向的矩形被覆各個格子狀配置。 選擇X方向、Y方向的任一個例如由如以下的評價基 準來決定。 甲、藉由XY台的安裝（上段、下段）’成爲上段台 的方向。 乙、矩形片數少即可完成者。 丙、動作距離（分割矩形的進行方向的邊）的總和小 -40- (37) (37)200302761 者（在圖38 ( B )爲Lx與Ly的總合比較）或分割矩形內 的點存在的矩形彼此之間的距離的總和小者（dx與dy的 總合比較）。 丁、模型化電掃描器與XY台的動作，藉由適切的動 作時間的簡易的模擬的選擇。 或者如圖3 8 ( C)所示，若單純地配置加工矩形的話 ，像開孔位置在端被二等分的加工矩形產生的情形是用以 避開無開孔位置的部分，與應配置有開孔位置的部分的中 心的加工矩形的中心重疊。 關於無法以在前述步驟42 1中的好的矩形被覆者（在 圖36爲B8、B9、B13 )，此處不確定矩形，在下一個圖 23的步驟408以最終剩餘的開孔位置的加工矩形配置處 理。 最終剩餘的開孔位置少的情形也能使用藉由台停止加 工的動作。即在圖22可見的基板的例子，以圓形包圍的 部分11 B爲少數群。對於這種位置被普遍配置於基板內的 情形，有少數群的每一個是使XY台朝停止位置動作、整 定，進行整定後照射者在加工時間的點有效的可能性。 此外，台停止加工在加工中速度爲〇，移動加工區域 間的時候視爲例如接近最大速度的高速的一定速度的速度 模型’台非停止加工的一部分可以相同的控制形態實施藉 由台停止加工的加工。 圖2的步驟1 1 2的開孔位置訪問順序最佳化以及Xγ 台等速動作速度最大化製程主要是組合以下的三個基本的 -41 - (38) (38)200302761 製程的處理。 A、 開孔位置訪問順序發生製程 B、 台動作速度算出製程 C、 開孔位置訪問順序改善製程 製程A (開孔位置訪問順序發生製程）是暫定地決定 一個如一邊直線狀地使XY台動作，一邊可進行加工的開 孔位置訪問順序（加工順序）的製程。本製程在本文以下 或圖面中如程式上的函數名，寫成initial_sequence()。 而且，製程B (台動作速度算出製程）是對某開孔位 置訪問順序，如一邊使XY台等速動作，一邊可加工的求 台動作速度的最大値的製程。本製程在本文以下或圖面中 寫成 find一speed ( ) 〇 而且，製程C (開孔位置訪問順序改善製程）是固定 台動作速度於某値，使加工矩形內的（包含電掃描器掃描 、整定等全部）總加工時間縮短，探索改善更佳的開孔位 置訪問順序的製程。儘可能連續重複使用本製程的處理 <C> (此處的記號<*>爲儘可能連續重複使用處理*的情形 的記號）在本文以下或圖面中寫成opt ()。 組合上述三製程，最單純如圖39所示僅藉由製程A 、B可進行本製程。 而且如圖40，也能依製程A、B、<C>、B的順序來 進行。 更佳爲如圖41所示，以製程 A、B、<<C>、B>來進 行也可以。即接著製程A、B儘可能重複『重複開孔位置 -42- (39) (39)200302761 訪問順序的改善以及台最大算出速度』。 再者，在後述如圖42所示，以製程A、B、<{C、B}> 來進行也可以。（此處的記號{*、**}爲表示*或**的任一 個的記號）。即接著製程A、B儘可能重複『開孔位置訪 問順序的改善以及台最大速度算出的任一個』。 或者，雖然若干要點不同但如圖43所示，也能以製 程A、<C’>、B。此處，製程CT爲相對於台動作速度被固 定於某値，台動作速度爲0的情形的最佳化，因此，爲不 伴隨著後述的實行可能性判定處理（feasibility ())。 具體上，高明地適用巡迴售貨員問題的解法（實際上爲藉 由哈密頓路徑長最小化問題的局部探索法的解法）。即在 使用例如Lin-Kernighan法等的巡迴售貨員問題的手法時 ，加入連結始點與終點的枝必須切斷的限制以解決問題的 手法。即若稍微使圖43具體的話，如圖44所示產生初期 路徑，解決附有連結路徑的始點與終點的枝必須切斷的限 制的巡迴售貨員問題，然後，可進行求出XY台速度的處 理。 以下，針對製程A、B、C詳細地說明。此外，在以 下的說明中也有僅將開孔位置記爲點的情形（當然說明上 無任何問題）。 製程A (開孔位置訪問順序發生製程）是χγ台等速 動作的情形的構想，如圖45所示而進行。即對χγ台動 作方向設定加工區域的尺寸以下的寬度h，進行長方形狀 的分割，針對該寬度h，將收容於該寬度h的開孔位置分 -43- (40) (40)200302761 類成與動作方向垂直的方向’在相鄰的寬若在某寬度內升 順分類的話，使在其接鄰的寬度內爲降順分類，可令使開 孔位置的訪問順序成爲蛇行狀的訪問順序爲初期路徑。 此外，如[應用數理計劃手冊](久保幹雄等編輯，朝 倉書店，2002 )所揭示的，這種路徑的產生方法一般稱爲 條（strip )法，惟在本發明中如以下所示決定分割寬度h 。即如圖45所示，令與矩形的XY台動作方向垂直的方 向的長度爲a，XY台動作方向的長度爲b，矩形內的孔數 爲N，想求的分割寬度爲h。此處，想由朝X方向的移動 量與γ方向的移動量相等的話無浪費的觀點，藉由求出 XY台進行方向的移動量的平均値和與XY台進行方向垂 直的方向的移動量的平均値的兩者一致的情形來決定最佳 分割寬度即像動作時間最小的分割寬度。 爲了統計的評價，假定矩形內的開孔位置爲隨機分布 〇 首先，XY台進行方向的移動量的平均在區間[〇、h] 內爲隨機發生的兩數的差的平均。該値考慮爲相對於滿足 〇 S X S h，0 $ y $ h的兩數X、y，兩數的差的絕對値 爲區域OSx^h，O^y^h中的z値的平均値，由下式計 算求出。 -44- (41) (41)200302761 (1/h2) H I [y I dxdy = h/3 ... ( 2) 另一方面，與χγ台進行方向垂直的方向的移動量的 平均因在寬度a的區間爲平均Nx ( h/b )個孔隨機排列時 的平均間隔，故變成下式 a/ ( Nx ( h/b ) ) =ab/(Nh) ... ( 3 ) 因此，兩者一致係如圖46所示，爲下式成立時。 h/3:ab/ ( Nh ) ... ( 4 ) 若計算此式的話變成下式 h =，（ 3ab/N) =/" ( 3/p ) ... ( 5 ) 此處，P爲密度。 製程B (台動作速度算出製程）是使用如圖47的方 法。 首先，在步驟9 3 0中進行後述的不可加工實行的台動 作速度算出處理（本處理在本文以下或圖面中寫成 max_speed())，求出不可實行的速度V2。 接著，在步驟9 3 2中進行後述的可加工實行的台動作 速度算出處理（本製程在本文以下或圖面中寫成 -45- (42) (42)200302761 min_speed())，求出可實行的速度vi。 在步驟934中以〇初期化反覆次數。 在步驟936中脫出條件放入反覆次數達ITE次的情形 或v2-Vl爲TOL以下的値的情形的環處理。 在步驟93 6中無法脫出的情形是在步驟93 8中求與 V2的中間値v。 而且，對速度v在步驟940中後述的實行可能性判定 處理（本製程在本文以下或圖面中寫成feasibility ()) ，調查以速度v是否可實行。 在步驟942中判定是否可實行，不可實行的情形在步 驟944中以v更新不可實行的速度V2，可實行的情形在步 驟946中以v更新可實行的速度。 此處，針對不可實行的台動作速度算出處理 max_speed ()、可實行的台動作速度算出處理min_speed ()以及實行可能性判定處理feasibility ()來說明。 首先，針對實行可能性判定處理feasibility ()來說 明。開孔加工的實行可能性如圖48所示，隨著XY台速 度的增大，由可實行的狀態移動到不可實行的狀態。因此 ，藉由如以下的方法算出實行可能性的程度。即如圖49 所示，對各點設時序（Timing)間框限制即最早開始時刻 與最遲完了時刻。此處，最早開始時刻是指可開始開孔用 的雷射照射的最早時序（Timing)，最遲完了時刻是指開 孔用的雷射照射必須完了的最遲時序（Timing)。具體上 如圖49，藉由XY台的反方向的動作令加工區域在γ方向 -46- (43) (43)200302761 以速度V在基板上等速遷移，在時刻0加工區域的上限位 置通過基準點Y = 0。令點P的Y座標値（絕對値）爲YP ，則最早開始時刻可以Yp/V，最遲完了時刻可以（Yp + D )/V求得。但是，D[理想上]表示加工區域的大小。 此處，雖然使用[理想上]的表現，但D的値實際上因 在上述驅動模式與現實的機械動作之間發生不吻合，故爲 了應付此點，D的値設定爲比加工區域的大小還小一些。 即在最早開始時刻側具有長度DF份的裕度（margin )，在最遲完了時刻側具有長度DB份（DF、DB爲0以 上）的裕度，進行對誤差的應付。此情形最早開始時刻可 以（Yp + DF) /V，最遲完了時刻可以（Yp + D-DB) /V求得 〇 以數字顯示例子。例如如圖50所示，令Yp = 300mm、 D = 50mm、V=1 00mm/秒，貝[J在無誤差應付用的裕度的情形 最早開始時刻爲 300/100 = 3.0秒，最遲完了時刻爲（ 3 00 + 50 ) /1 00 = 3.5秒。因此，某開孔位置的加工完了（即 照射完了）時刻可判定爲例如3.2秒即可，3.7秒的話不 可。 現在，令第i個訪問的點爲Pi，最初訪問的點爲P!的 照射開始時刻爲P:的最早開始時刻。各點的照射完了時 刻可藉由對照射開始時刻加入照射時間的時刻求出。200302761 (1) 发明. Description of the invention [Technical field to which the invention belongs] The present invention relates to a low-speed positioning means that can move a machining execution position on a workpiece in a wide range while coordinating actions, and has a side length DX in the X direction and a Y direction A narrow-range movement within a processing area of a predetermined size of a side length DY is a possible high-speed positioning means, and a processing planning method and apparatus for processing a plurality of processing positions scattered on a workpiece while processing are determined by the aforementioned processing planning method A processing method for processing, a processing device including the processing planning device, a computer program for implementing the processing planning method or implementing the processing planning device, and a computer-readable recording medium storing the computer program. [Prior Art] In recent years, with the demand for miniaturization or high-density mounting of electronic equipment, a multilayer printed wiring board having a plurality of printed wiring boards stacked thereon has been gradually provided. In such a multilayer printed wiring board, a conductive layer called a through hole or a via hole is formed in each of these substrates for electrical connection between conductive layers formed on top and bottom of the printed wiring board. hole. Then, a conductive film is formed inside these holes to connect the conductive layers of the printed wiring boards. The hole formed in the printed wiring board has been miniaturized in accordance with the recent miniaturization or high functionality of the printed wiring board, and has a diameter of 0.1 mm or less. In order to form such a small-diameter hole with high accuracy, a pulsed laser beam (2) (2) 200302761 is used. The general structure of a laser holer using a conventional pulsed laser is shown in Figure 1. Show. This configuration example includes a first electrical scan including a rotating mirror 23 that is irradiated by a laser oscillator (not shown) and scans a pulsed laser beam 20 in a predetermined direction (in the direction perpendicular to the paper surface in FIG. 1). Scanner (galvano scanner) 22, which includes scanning by the first electrical scanner 22 in a direction perpendicular to the scanning direction by the first electrical scanner 22 (a direction parallel to the paper surface in FIG. 1). The second electrical scanner 24 of the rotating mirror 25 of the laser beam scanned perpendicular to the paper surface, and the laser beam scanned in both directions are fixed to the XY stage by the first and second electrical scanners 22 and 24. The surface of the processing object (referred to as a work) 10 on the substrate 12 is deflected perpendicular to the f-0 lens 26 for irradiation. In this way, by using the first and second galvano-scanners 22 and 24 to reflect the laser beam 20 to the rotating mirrors 23 and 25 at the front end of the scanner, the direction of travel can be arbitrarily changed. Here, since the rotating mirrors 23 and 25 are lightweight, high-speed positioning is possible. The laser light deflected by the aforementioned electrical scanners 22 and 24 passes through the f-0 lens 26 and is collected on the workpiece 10. Since the f-0 lens 26 is generally expensive, its size is limited. Therefore, the beam irradiation range (referred to as the area) in a certain timing (Timing) is limited to an angle of about several tens mm. The rectangular shape is narrower than the size of a normal workpiece. Therefore, a wide range of positioning is possible by transferring the workpiece 10 using the X Y stage 12. However, due to the large weight of the XY stage 12, the time required for positioning is large. 0 -6-(3) (3) 200302761 So, the laser drilling machine uses a high-speed narrow-range positioning device with an electric scanner and a low-speed wide The two positioning devices of the XY stage of the range positioning device perform high-speed and wide-range hole positioning. Therefore, from the viewpoint of positioning, the motion patterns (control patterns) can be largely classified into the following two. (1) The separate control of the electric scanner and the XY stage, that is, laser processing by laser light scanning in the processing area in a state where the XY stage is stopped, and laser processing by a χγ stage workpiece are performed in series. The form of conveying and processing the entire substrate is generally called step and repeat. Hereinafter, it is also referred to as low-speed positioning means stop processing or table stop processing. (2) Coordinated control of the electric scanner and the XY stage (also referred to as synchronous control) The form of scanning the laser light by the electric scanner and the conveyance of the workpiece through the χγ stage in parallel is usually called coordinated control Or synchronous control. Hereinafter, it is also referred to as non-stop processing or stage non-stop processing by low-speed positioning means. The aforementioned stepping and repetition are disclosed in, for example, US Patent No. 6087625, and the coordinated control is disclosed in Japanese Patent Application Laid-Open No. 3-266006. However, U.S. Patent No. 6087625 relates to stepping and repetition, and does not include a machining plan in coordinated control as an object. In addition, "Japanese Patent Application Laid-Open No. 3-266006 is a coordinated control of a coarse motion stage and a micro motion stage" is completely different from the subject of this case. (4) (4) 200302761 [Summary of the Invention] The object of the present invention is to create a solution to solve the conventional problems, and effectively plan to move a laser on a workpiece in a narrow range in the pre-processing stage. The irradiation position is a possible high-speed positioning means, and a wide-range movement is possible as a low-speed positioning means. The present invention is a low-speed positioning method that makes it possible to move a machining execution position on a workpiece in a wide range while coordinating actions, and a narrow range in a machining area of a predetermined size having a side length DX in the X direction and a side length DY in the Y direction. Ground movement is a possible high-speed positioning method. The processing planning method when processing a plurality of processing positions scattered on the workpiece, including: by the low-speed positioning means during processing such as X or Y, the side length of the direction Z is DZ or less. Processing rectangle division process for dividing the processing position; for each processing rectangle determined by the processing rectangle division process, the total processing time in the rectangle is minimized, and the processing position access order and processing speed of the low-speed positioning means in the processing rectangle are determined. Position access sequence and operation speed optimization process of low-speed positioning means; and determine the low-speed positioning means based on at least one of the processing speed of the processing rectangle position and the processing speed of the entrance and exit of the processing rectangle determined by the two processes Low-speed positioning method of the motion path optimization process of the motion path In order to solve the aforementioned problems. In addition, the rectangular division process (means) for this processing is performed by simple division in the X direction or the Y direction. In addition, after the simple division in the X direction or the γ direction is performed, the processing rectangle that does not exist in the processing position is removed, and the operation position of the low-speed positioning means is limited to the position with the processing position. -(5) (5) 200302761 Effective coordinated control. Furthermore, if the length of the side of the segmented processing rectangle in the stop direction Z of the low-speed positioning means is shorter than a side of a predetermined dimension in the direction, the side length in the stop direction is adjusted without changing the number of divisions. Eliminate machining rectangles with small sides in the stop direction. Moreover, the process (means) for dividing the rectangular shape is performed by the least number of divisions in the X direction or the Y direction. In addition, the rectangular cutting process (means) for this process is to test the X-direction and Y-direction divisions on both sides, and then select the best one (for example, the one with a low-speed positioning means with a short operating distance) to achieve high-speed processing to achieve effective Coordinated control. In addition, after the processing rectangle is divided, in each processing rectangle, if the processing position along the operating direction of the low-speed positioning means has a processing area having a predetermined size without the operating direction of the low-speed positioning means, the side length and speed required for the processing area change. For distances and above, the rectangle is bisected. Then, after dividing the processing rectangle, the processing position is set to the center of each processing rectangle in a direction perpendicular to the operation direction of the low-speed positioning means, and the arrangement of the processing rectangle is finely adjusted. In addition, after making the processing rectangle division process (means) to perform the X-direction division and the Y-direction division on both sides to determine the processing rectangle, the combination of these (for example, the short-motion positioning means with a short operating distance) is used to process the entire rectangular division. High-speed processing enables effective coordinated control. In addition, the processing rectangle division process (means) is used to consider the distribution pattern and density of the processing positions. By considering the dense rectangular division of the processing position (6) (6) 200302761, the sparse positions are at Sparse locations, dense locations are specified in dense locations with a specific rectangle. Processing is performed at a fast speed at a sparse location and at a slow speed at a dense location, and high-speed processing can achieve effective coordinated control. Furthermore, the process (means) for dividing the processing rectangle that takes into account the distribution pattern and density of the machining positions includes: identifying the positions where the machining positions are dense and determining the arrangement of the machining rectangles (means); identifying the positions where the machining positions are the majority, Determine the manufacturing process (means) of the configuration of the processing rectangle; and determine the manufacturing process (means) of the processing rectangle for the processing position that is not determined. In addition, the manufacturing process (means) for arranging the identified dense locations and the processing rectangles at most locations includes: dividing the processing area on the workpiece into a grid shape, and identifying the density of the processing location in each grid or the authenticity of the processing location. Process (means): a process (means) to form a plurality of grid group rectangles based on the identified grid configuration: and a process (means) to cover the rectangles of each grid group and configure a processing rectangle. Further, the grid group rectangle has a grid that is identified as a certain true in the grid row corresponding to the outer periphery of the rectangle, and the grids in the grid sub-rows corresponding to the outer periphery of the rectangle are all recognized as pseudo grids. Furthermore, in a process (means) for identifying a processing position where there are a large number of positions and determining the arrangement of the processing rectangle, the expansion of the processing position recorded by repeating the data itself as a pattern is recognized as the majority position. In addition, when the processing rectangles are arranged to cover the rectangles of the grid groups, the grid group rectangles that do not satisfy the set size are excluded, and the processing rectangles are determined -10- (7) (7) 200302761. Furthermore, in this process (means) for covering the rectangular shapes of the grid groups and arranging the processing rectangles, the linear or grid-like arrangement of the grid group rectangles is recognized, and the processing rectangles are determined. In addition, when there are few remaining machining positions, the low-speed positioning means is stopped, and the low-speed positioning means that moves the high-speed positioning means is stopped to make the processing effective. Furthermore, the same control system as that of non-stop processing by the low-speed positioning means is executed to stop the processing by the low-speed positioning means, and the design is facilitated. Further, the process (means) for optimizing the access sequence of the processing position and the operating speed of the low-speed positioning means are to be obtained while optimizing the access order of moving, positioning, and processing by the high-speed positioning means to a possible processing position. This low-speed positioning method is a process (means) that optimizes the processing position access sequence of the maximum speed of constant speed and the constant-speed operation speed of the low-speed positioning method. Furthermore, the process (means) for optimizing the access sequence of the processing position and the operation speed of the low-speed positioning means are optimized after the process (means) for optimizing the access sequence of the processing position and the constant-speed operation of the low-speed positioning means. According to the distribution density of the processing positions, the speed model determines the process of determining the acceleration / deceleration model that smoothly changes the operation speed. In addition, in the process of optimizing the access sequence of the processing position and maximizing the constant-speed operation speed of the low-speed positioning means, the processing timing (Timing) defining each processing position is determined based on the operating speed of the low-speed positioning means -11-( 8) (8) 200302761 The time frame between the earliest start time and the latest finish time. Furthermore, in the process of optimizing the access sequence of the machining positions and maximizing the constant-speed operation speed of the low-speed positioning means, the movement length between the processing positions is determined based on the amount of movement between the machining positions and the operating speed of the low-speed positioning means. Then, the time frame of each processing position is set to the side length DZ of the low-speed positioning means operating direction z using a processing area of a predetermined size of X or γ, the operating speed V of the low-speed positioning means, and the processing start time TS of the processing area. The distance PZ from the processing position P to the nearest edge of the processing area of the predetermined size is determined by the earliest start time of TS + PZ / V and the latest finish time by TS + (PZ + DZ) / V. Then, the time frame of each processing position is set to the side length DZ of the low-speed positioning means operating direction Z using the processing area of a predetermined size of X or Y, the operating speed V of the low-speed positioning means, and the processing start time TS of the processing area. The distance PZ from the processing position P to the nearest edge of the processing area of a predetermined size is smaller than DZ by a predetermined positive or 0 値 DZF, DZB, the earliest start time is TS + (PZ + DZF) / V, and the latest finish time It is determined as TS + (PZ + DZ-DZB) / V. In addition, in the process of optimizing the access sequence of the processing position and maximizing the constant-speed operation speed of the low-speed positioning means, after the processing in a previous processing position is completed, the time when the movement between the processing positions is completed is The earliest start time is still early. If the earliest start time is after the earliest start time, the processing start time for each processing position is determined as the time when the movement is completed. -12- (9) 200302761 Moreover, the process (means) for optimizing the access sequence of the machining position and maximizing the speed of the isokinetic operation in the low-speed stage is the process (means) for the machining position access sequence that causes the machining positions to occur sequentially; The machining position access sequence calculates the constant speed operation of the low speed positioning means. The low speed positioning means operation speed calculation process (means) constitutes. In addition, the process (means) for optimizing the processing position access sequence and maximizing the speed of the isokinetic operation in the low-speed stage are all including the process for generating the processing position access sequence that generates the access sequence (the method determines the processing position access sequence to calculate the low-speed positioning means). Constant speed low-speed positioning means operation speed calculation process (means); and fixed-position means operation speed to improve the processing position access order processing process improvement means (means). Moreover, the processing position access order is optimized and low speed The step (means) for maximizing the isokinetic operation speed is a process (means) that generates the machining position access sequence that causes the machining positions to occur sequentially. The work position access sequence is used to calculate the isokinetic operation speed of the low-speed positioning means. ): Repeat the fixed low-speed positioning means action speed several times, improve the processing position access sequence to improve the processing position access sequence improvement process (means) and the low-speed fixed action speed calculation process (means) repeatedly improve the process (means into. Also, make The processing location interview The process (means) of order optimization and constant-speed action speed maximization at low speeds are made by the process (means) of sequentially accessing the processing positions of the processing positions; positioning manual positioning and manual positioning of the fixed speeds); The speed is fixed at a low speed, and the position is set by a fixed speed. The low speed is set to the interrogation sequence.), And the positioning is not accompanied by a -13- (10) (10) 200302761. The determination of the possibility of processing is repeated. Processing position access sequence improvement process (means) without judgment. Processing position access sequence improvement process (means); and fixed processing position access sequence to calculate low speed positioning means's constant speed operation speed. Low speed positioning means operation speed calculation process (means). Make up. In addition, the processing sequence formed by the processing position access sequence generating process (means) is to divide the processing rectangle by setting the width of the operation direction of the low-speed positioning means, and classify the processing positions accommodated in the width within the width. If the operation direction is vertical, if the ascending order is classified in a certain width, the descending order is classified in the adjacent width, so that the access path of the processing position is in a meandering order. Further, the division width when the width is divided into band-shaped rectangles for the operation direction of the low-speed positioning means is set based on the density of the machining positions in the machining rectangle. In addition, the repetitive improvement process (means) is set to 値 V between the operation speed of the high-speed low-speed positioning means and the possible low-speed operation speed that are impossible for processing. First, it is determined whether it can be implemented and the feasible situation is updated with V. Low-speed operation speed, if it is not feasible, repeatedly execute the machining position access sequence improvement process (means) for V, and it becomes feasible. Update the low-speed operation speed with V, and finally it will not become feasible until the repetition ratio is set. The process (means) of repeating the process of updating the high-speed operation speed with V until the number of times is high or the difference between the high-speed operation speed and the low-speed operation speed is smaller than the set value 値. In addition, the repetitive improvement process (means) is performed to implement the low-speed positioning means operation speed V that can be -14- (11) (11) 200302761, and the processing position access sequence is improved to improve the process (means) to the number of repetitions. Until the setting times are too many, or the total processing time before and after the process order access method (means) cannot be shortened, or when the process site access order improves the process (means), the speed improvement ratio of the low-speed positioning means is improved. Up to this point, the process (means) of renewing the processing speed that can be performed when the solution disappeared was repeated. In addition, the process (means) for improving the access sequence of the machining position is applicable to the solution of the asymmetric Hamiltonian path length minimization problem with a time frame or the symmetric Hamiltonian path length minimization problem with a time frame. Moreover, the method of minimizing the problem of symmetric or asymmetrical Hamiltonian path length with a time frame uses the Or-Opt method. In addition, the process (means) for improving the access sequence of the processing position and the operating speed of the low-speed positioning means are calculated based on the difference between the processing completion time and the latest completion time of all processing positions. . In addition, the processing order access improvement process (means) of the processing position and the operation speed calculation process (method) of the low-speed positioning means are made possible. If the processing time of all the processing positions is completed, the latest time sequence (Timing) ) Is determined to be possible if it is not too late, otherwise it is determined to be impossible. Furthermore, the improvement solution in the processing order access improvement process (means) is determined based on at least one of the processing completion time of the processing position visited last in the processing rectangle and the possibility of execution of the processing. -15- (12) 200302761 Furthermore, the process of optimizing the access order of the processing position and the speed of the low-speed positioning means (means) is to first determine the order of processing positions in the rectangle, and secondly, optimize the path in the determined path. Constant speed action for low speed positioning. Furthermore, the speed model determines the process (means) to increase the speed of the low-speed positioning operation speed in the part where the distribution of the processing positions in the rectangle is sparse. Furthermore, as a result of optimizing the processing sequence (means) of the access order of the processing positions and the speed of the low-speed positioning means, if the processing completion time of all processing positions is earlier than the latest completion time, the processing completion time is in all positions. The range that is earlier than the latest finish time is shifted from all the processing positions in the direction of the same time. In addition, the process () of optimizing the operation path of the low-speed positioning means is to minimize the sum of the time between moving processing rectangles. In addition, the optimization process () for optimizing the operation path of the low-speed positioning means is to solve the problem that the city is a position corresponding to the entrance and exit of each processing rectangle. Salesman Problem or Hamilton Path Length Minimize Problem Course. Furthermore, the optimization process () of the operation path of the low-speed positioning means is to determine the Q of each rectangle after optimizing the access sequence of the processing rectangles. In addition, the process of optimizing the operation path of the low-speed positioning means (processing in the action visit means processing means operation, the entry and exit means of each prohibited means at the beginning of the means means-16- (13) (13) 200302761) is After determining the entrance and exit of each rectangle, the access order of the rectangles is optimized. The present invention is a low-speed positioning method that makes it possible to move a machining execution position on a workpiece in a wide range while coordinating actions, and moves in a narrow range within a machining area of a predetermined size having a side length DX in the X direction and a side length DY in the Y direction. As a possible localization speed positioning method, the processing planning method when processing a plurality of processing positions scattered on the workpiece includes a full machining position access order optimization process (means to optimize the access order of the processing positions of the entire workpiece) ); And the low-speed positioning means action model determines the process (means) for the processing position access order determined by the full-process position access order optimization process (means), and the low-speed positioning means action model is determined. In addition, the low-speed positioning means operation model is used to determine the process (means). First, the processing positions are distinguished by rectangular groups of the same size as the predetermined size rectangle, and then the low-speed positioning means corresponding to the processing positions of each group are controlled. Low-speed positioning means is possible to move the machining execution position on the workpiece over a wide range, and high-speed positioning means is possible to move within a narrow range within a processing area of a predetermined size having a side length DX in the X direction and a side length DY in the Y direction. The processing planning method when processing a plurality of processing positions scattered on the workpiece while one side includes: by using a low-speed positioning means such as X or Y during processing, the side length of the direction z is less than DZ, and the processing rectangle is divided into processing positions. A division process (means); a process (means) that optimizes the access order of processing positions for each processing rectangle that is determined by the processing rectangle division process (means); and -17- (14) 200302761 Work location access order determined by optimization process (means) Positioning means for low speed operation of determining the motion model models positioning means determines a low-speed process (means). The present invention provides a processing method 'characterized in that processing is determined by the processing planning method described in any one of the foregoing. Furthermore, a computer program for implementing the processing planning method described in any one of the above is provided. Further, a processing apparatus including the processing planning apparatus described in any one of the foregoing is provided. Furthermore, a computer program for realizing the processing planning device according to any one of the foregoing is provided. Furthermore, a computer-readable recording medium having the aforementioned computer program recorded is provided. [Embodiment] Hereinafter, an embodiment of the present invention in which a hole is opened by a laser holer including an XY stage (low-speed positioning means) and an electric scanner (high-speed positioning means) will be described in detail with reference to the drawings. First, a brief description of the concept that forms the basis of the invention. In Japanese Patent Application Laid-Open No. 200 1 -33 1 550, which has been proposed by the inventor, it is proposed to implement stepwise and repetitive actions (ie, low-speed positioning means (sets) to stop processing) for high-speed implementation, and to plan χγ sets before processing (also (Referred to as a stage only), a stop position access sequence, and a scanning track of an electric scanner. This is the optimizer for the following three items. -18- (15) (15) 200302761 < 1 > Dividing the processing area if it is the hole position (processing position) (if it is the range of the irradiated laser beam, the shape is not limited to a rectangle) (reduction in the number of XY stage operations) < 2 > Access order of opening positions in the processing area (reduction of scanner operation time) < 3 > Access sequence of processing area (reduction of XY table movement amount) The present invention is to expand the above-mentioned concept for low-speed positioning means to stop processing (here, table stop processing) to low-speed positioning means non-stop processing (this The processing of the non-stop machine), the following four items are used to optimize the 〇 (1), the processing rectangle division of the hole position (limitation of the XY table operation position) (2), access to the location of the hole in the processing rectangle Sequence (reduction of scanner operating time) (3), XY stage speed (including acceleration / deceleration model) when processing rectangles pass (reduction in XY stage waiting time) (4), access sequence of processing rectangles, and entrance and exit of processing rectangles ( Decrease in the amount of movement of the XY table) < 2 > The optimization is divided into the optimizations of the two items (2) and (3). Here, the [processing rectangle] refers to a region that becomes a unit of processing when the XY stage is processed while being straight in one direction without stopping, and the processing region extends in a shape (rectangular shape) in the XY stage operation direction. In addition, although the low-speed positioning method here uses XY stage and high-speed positioning hand -19- (16) (16) 200302761 to explain the use of an electric scanner as an example, the types of low-speed positioning means or high-speed positioning means are not limited. herein. Regarding the non-stop processing (2), the order of accessing the hole positions in the machining rectangle can be considered as the optimal sequence is different due to the speed of the XY table in (3). Conversely, the maximum speed of the XY stage speed of (3) will vary depending on the order of access to the hole positions in the machining rectangle of (2). Therefore, since (2) and (3) are closely related among the above four items, it is better to optimize together rather than independently. Therefore, as shown in FIG. 2, it is divided into three processes which make the whole independent. A. Rectangular cutting process of the hole position (step 1 00) B. Access order of the hole position (within each processing rectangle) and the XY stage motion speed optimization process (step 110) C. The XY stage has the best motion path Process (step 120) The process of optimizing the rectangular segmentation at the hole position of step 100 is shown in FIG. 3. This process is performed in a direction perpendicular to the XY stage (actually the direction of the processing area migration on the substrate, and the stage moves in the opposite direction, but as shown to the arrow writer for easy understanding, write the stage progress direction). A process that divides a hole position into a rectangle (processing rectangle) that is parallel to the substrate arrangement direction with a length equal to or less than the side length of the processing area. The order of accessing the hole positions in each processing rectangle in step 110 and the thoughts of the XY stage operation speed optimization process are shown in FIG. 4. This process is related to each processing rectangle determined in step 100, and minimizes the difference between the processing time of the processing rectangle, that is, the processing completion time (laser irradiation) of the last processing position and the processing start time of the first processing position. XY stage-20- (17) (17) 200302761 operating speed (not a certain speed is also possible), while optimizing the processing order of the opening position access sequence in the rectangle. The problem of the size of the processing area when the βρ XY stage operates at a constant speed. 'Timing' that can be performed even if laser irradiation is started at each opening position, and Timing (which is called time frame) must be completed. limit). For example, if the speed of the χγ stage is extremely slow, although the irradiation of all the opening positions is not caused too late, the processing time becomes longer. On the other hand, if the speed of the XY stage is too fast, it may happen that the positioning due to scanning by the electric scanner is too late. In addition, the scanning time of the scanner is affected by the scanning distance of each scan and the access sequence of the hole positions. From the above, it is required to access the order of opening positions where scanning, positioning, and irradiation with a scanner are possible (hereinafter, the irradiation timing (Timing) that can be implemented for all the opening positions in the processing rectangle is determined to be possible). And the maximum speed of XY stage. In this process, there are countless necessary factors such as the order of access to the hole positions, the timing of the irradiation of each hole position (Timing), the XY stage speed, and the acceleration / deceleration model. It is an extremely difficult process, so it is divided into the following Two processes. B 1. The process of determining the sequence of accessing the hole positions in the processing rectangle with the largest operating speed when the XY stage is operated at a constant speed (step 1 1 2). The following describes the process of optimizing the access sequence of the hole position (machining position) and the process of maximizing the constant speed of the XY stage (low-speed positioning means). B2. For the above determined opening positions, the distribution of the opening positions is dense. -21-(18) (18) 200302761 degrees determine the process of the XY stage acceleration / deceleration model that smoothly changes the operating speed (step 1 1 4) . In addition, from the processing accuracy, the pattern of the opening position of the substrate, the design of the control system used, etc., it can be judged that the stage makes the stage operator better at a constant speed. Steps 1 to 14 are skipped. The idea of the process of optimizing the operation path of the XY stage in step 120 is shown in FIG. 5. This process is based on the position of the processing rectangle determined in step 100 and the operation speed of the X / Y table of the processing rectangle inlet and outlet determined in step 110 as the basis to optimize the X / Y table's motion path (the processing rectangle access order of the processing area and Process area migration direction on each processing rectangle). Hereinafter, each process will be described in detail. In the aforementioned step 100, the optimization process for dividing the processing rectangle at the opening position is to divide the opening by processing the rectangle parallel to the arrangement direction of the substrate by using a direction perpendicular to the direction of the χγ stage to be less than the width of the processing area. Location process. The method of implementing this process is considered as the following five methods. 1. The method of simple division in the horizontal or vertical direction 2. The method of removing the position where the hole does not exist by the horizontal or vertical division 3. About the above 1 and 2, try the horizontal and vertical directions on both sides. Choose the best method 4. Regarding the above 1 and 2, the two sides try to determine the processing rectangle in the horizontal and vertical directions, and then use the combination of these processing rectangles to divide the whole method by the processing rectangle. 5. Consider the same pattern and density. Machining Rectangular Segmentation-22- (19) (19) 200302761 Methods are described below for each method. 1. The simple division method in the horizontal or vertical direction is shown in Fig. 6. In the X direction (horizontal direction) or the γ direction (vertical direction), the progress direction of the XY stage (horizontal direction in the figure) is determined in advance, and the stage moves. The size of one side of the processing rectangle in the vertical direction (Y direction) is taken as 値 below the size of the processing area, and the size of the side in the stage direction (X direction) is taken from the maximum 値 of the opening position minus the minimum.値 加工, a method of sequentially processing rectangular shapes arranged at the end of the substrate without any space. Although it may not be the best segmentation, as shown in Fig. 7, since the opening position of the substrate is generally enlarged to the whole, there is no solution even if you try to optimize, so this method is also an effective method. Depending on the situation, you can join the following efforts. a. How to deal with the processing rectangle that does not exist in the hole position. In the case of simple division in the horizontal or vertical direction, there may be cases where the processing rectangle has no hole position at all. Since such a processing rectangle is unnecessary, a removal process is performed. b. Setting method of the size of one side in the direction perpendicular to the direction of XY stage. Suppose that the XY stage moves in the horizontal (X) direction. That is, let the direction of the XY stage be horizontal (X), and the direction perpendicular to the direction of the XY stage (stop -23- (20) (20) 200302761 only) be vertical (Y). Let the size of the processing area be DM during the processing direction, and the vertical (Υ) width of the DS ′ substrate during processing (the largest and smallest difference in the γ coordinate of the hole position) be τ. In this method, the number of pieces of the processed rectangle becomes [(T_a) / DS] + 1 °. Here 'is a Gauss symbol' α which represents the largest integer not exceeding the number A, which means that it is very small compared with T and DS. Positive zero near zero. (^ Is the coping method for the case where T / DS is just divisible). For example, in the case of T = 400 mm and DS = 50 mm, the number of pieces of the processed rectangle is 10 pieces. For the case of T = 400 mm and DS = 40 mm, α = 0 · 0 0 1 mm is still 10 pieces. Here, the length in the stopping direction of the processing rectangle does not need to be the same as the size D S of the processing area, and it is preferable that one side of the processing rectangle is the same as the size of the processing area or is small. Therefore, the vertical length T of the substrate divided by the number of pieces of the processing rectangle [(T-a) / DS] + 1 is the vertical length of each processing rectangle. By this means, for example, in a substrate of T = 460 mm and D = 50 mm, division by 460/10 = 46 mm is performed, so that the phenomenon that the processing rectangle with a width of 10 mm at the end of the division of 50 mm remains is eliminated. In order to achieve mechanical limitations or high-precision machining, since the XY stage has a limit on the operating speed, it is necessary to avoid the state in which the XY stage moves at a speed close to the limit speed in the processing of a narrow processing rectangle. However, due to this method, The width of the processing rectangle is uniformized, so this can be avoided. c. Deletion of the position without openings and division of the processing rectangle In the example shown in Figure 8 (A), if the opening position is searched from left to right in the processing rectangle, for example, there is no opening position at all. Position -24- (21) (21) 200302761 exists. Even if such a position is not stopped, the table operation is wasted. Therefore, as shown in FIG. 8 (B), it is desired to divide and process the rectangle. The basis for dividing the processed rectangle is, for example, as shown below. A. There is no point at the left or right end of the processing rectangle. B. When the processing rectangle is turned in the right direction to find the hole position, there is no point above [the width of the processing area DM + the distance α required for speed change] presence. Here, the reason why the division criterion is the position at which no point is greater than DM + α is as follows. That is, as shown in FIG. 9 (A), when the block at the opening position is bisected, it is investigated whether the processing rectangle can be divided into two left and right processing rectangles. Regarding the processing end position of the left processing rectangle and the processing start position of the right processing rectangle, the processing end position of the left processing rectangle needs to be located to the left of the processing start position of the processing rectangle on the right, but the positions are as shown in Figure 9 (B) The figure shows the center position when writing the processing areas adjacent to the left and right of the processing rectangle. Therefore, even if it is the lowest, the processing rectangles that need to be arranged in the progress direction have a space of DM or more. In addition, when moving from the processing end position of the left processing rectangle to the processing start position of the right processing rectangle, as shown in FIG. 9 (C), it is necessary to change the operating speed in the processing rectangle. Therefore, a little movement distance α is required (α 値 is appropriately determined by the characteristics of the stage used). In addition, the minimum interval between machining rectangles needs to be above DM + α. d. The method of fine adjustment of the position of the processing rectangle is perpendicular to the direction of the processing rectangle (here, the Y direction). -25- (22) (22) 200302761 The configuration is such that the positions of the holes are collected in the center of the processing rectangle. Staggered up and down. That is, in a case where a processing rectangle has a set of opening positions P = {Pi, pi + 1 ... pj (i $ j) and the rectangle is divided, the center of the Y coordinate 値 of the point belonging to P is enlarged to expand the center (Average of the largest 値 and the smallest 値) is the center position in the longitudinal direction of the processing rectangle, and the movement of the galvano-scanner can be concentrated in the center. Or you can use the center of gravity (when the center of gravity is up or down, all points are not beyond the limit of the processing rectangle) instead of the enlarged center. 2. Horizontal or vertical segmentation, the method of removing the position where the hole does not exist. As shown in Fig. 10, the position without the hole is removed to perform rectangular segmentation. Specifically, if the situation where the XY table moves in the horizontal direction, the position of the rectangle processed by this method is not to be a different configuration as shown in Fig. 1 (A), but to be neat as shown in Fig. 11 (B). To a consistent configuration. Therefore, only the Y coordinate information of the opening position is used to determine the arrangement of the machining rectangle. Specifically, the problem (X) is defined as follows (referred to as SPLIT), and the following (Y) is used to deal with problems using the SPLIT solution. Problem (X) (SPLIT): η number of data to be sorted in ascending order A data set A = {a 1, a 2… a η}, segmentation (segmentation refers to any i, j of i and j, Bi Π Bj = 0 (empty set) holds, and m U wBi = A is the minimum number of coprime partial sets (two sets of coprime refers to the elements in the set that are not the same) Β !, B2 … Bm. However, the elements of Bk are classified, and the elements at the end of Bk must be smaller than the elements of the head of B k +1, and the set of parts b k == {a s ... a. } About the front and end elements satisfies | ac-as | < = DS (DS is a fixed size indicating the size of the scanned -26- (23) (23) 200302761 area). Processing (Y): < 1 > Only the Y-coordinate 値 of the opening position of the entire substrate is taken out (the case where the vertical axis is divided. The case where the horizontal axis is divided is the X-coordinate 値) are classified without duplicates, and are used as elements of the set A (A = {ai ... an}, ai < ai + i). < 2 > Let A be the input of the problem SPLIT, and obtain a partial set B 1… B m. < 3 > Based on each set of parts B! ... Bm, the opening positions are divided into m groups. The problem of SPLIT's solution is specifically to capture one of the network optimization problems when weighting each side of the graph by finding the shortest path from the given node to the shortest path to all nodes, and apply a known solution. The whole is handled as shown in Figure 12. That is, first at step 200, as shown in FIG. 13 (A), only the Y coordinate of the opening position 位置 is repeated and classified without being repeated, as the elements of the set A = {a :, a2 ... an} (here, classification Means i for 1 gi S n-1, making ai < ai + 1 and rearrange). Furthermore, as shown in FIG. 13 (B), let the node set V = {v!, V2,..., Let the element Vi of V be attached to ai. Moreover, let the two nodes (Vi, L) (i < j) is a neighbor to the following cases. ai + D ^ aj or ai + D g aj-i and ai + D < aj That is, Vj adjacent to Vi is a command that corresponds to Vi's 为 ai is a certain processing rectangle-27- (24) (24) 200302761 SQ1's lower Y coordinate, let 对应 ^ corresponding to Vj be the lower γ coordinate The processing rectangle can be a processing rectangle adjacent to s Q 1 in the substrate. Furthermore, two adjacent nodes (Vi, Vj) (i < j) The weight of the pair is to make all 1 (weight refers to the length (cost) of the branch between two points). And 'Let the starting point be Vi (the point with the smallest Y-coordinate), and the candidate set of ending points is V e = {vk… vn} (but k refers to the investigation from k = 1 according to Shengshun and become the first of ak + D ^ an , That is, if the processing rectangle of the end point candidate set is at the lower position, the point with the largest 値 an in the Y coordinate is included). From the above, a directed network with v 1 as the starting point can be obtained in step 202. For the obtained network, in step 204, a known method such as the method of Dijkstra is used to solve the shortest path problem. Then, as shown in Fig. 3 (B), the shortest path and the shortest path length from the starting point to each node can be obtained. In addition, Dijkstra's method and so on are indeed methods that can obtain the shortest path (here, the least processed rectangle). Next, among the nodes in the candidate end point selection set in step 206, the shortest path length shows the smallest node. Next, in step 208, if the processing rectangle division corresponding to the Y coordinate of the node 値 below the shortest path toward the node is used, as shown in Fig. 14 (C), the desired processing rectangle division can be obtained. Here, the shortest path length becomes (the number of processing rectangles -1). Moreover, although the above method of setting the longitudinal position of the processing rectangle is for the purpose of minimizing the number of processed rectangles, the number of processed rectangles is the same as shown in Figure-28- (25) (25) 200302761 1 5 (A). There are still holes cut off at both ends by processing the rectangular covering (upper and lower figures). Therefore, among the solutions with the least number of processed rectangular pieces, it is better to choose the position where the opening is the most unsolvable. In order to achieve this, for the smallest number of processed rectangular pieces, the above-mentioned gap (gap) 値 is larger than the specified 位于 is located in the processing rectangle. 'If the gap 値 is added, choose the distance 値The sum is the best. Or another implementation method, adding the space 加工 between the processing rectangles, that is, the space 要素 between the final elements of the processing rectangle and the first elements of the processing rectangle on the first segment, and the sum of the distances 値 may be the largest. In addition, in this technique, various methods of processing rectangles of the above 1, C, and d can be added. Regarding the processing rectangle division by the methods 1 and 2 described above, the processing for optimizing the processing rectangle operation path in steps 1 to 20 in FIG. 2 is simple. That is, because the processing rectangles are arranged in the same section, the XY stage is in the horizontal direction. If the processing rectangles located below the vertical position in the substrate are in order, the movement path of the XY stage is sequentially (as already explained, the processing area is correct. The migration path) is preferably a meandering path from the bottom to the top of the substrate. The overall processing in this case is shown in Fig. 16. That is, in step 220, a rectangular segmentation is performed in which the segments are consistent by the method of 1, or 2. Next, in step 222, the operation path of the χγ stage is determined so that the downward direction of the substrate becomes a meandering shape. Next, in step 224, the order of access to the hole positions of each processing rectangle is optimized and the operation speed of the XY table is maximized. -29- (26) (26) 200302761 In addition, steps 222 and 224 may be reversed. 3. Regarding the above 1 and 2, after the two sides test the horizontal direction and the vertical direction, the method of selecting the best one is shown in Fig. 17 in this method. Try to process the horizontal and vertical processing rectangles to divide. Calculate each action distance, and choose the smaller one. In Fig. 17, the processing rectangle is divided in the horizontal direction, and the operating distance is short. Therefore, the processing rectangle in the horizontal direction is selected. Here, although the one with a shorter operating distance is selected, the selection criterion may be a method that takes into consideration the sparseness of the hole positions, for example. That is, as shown in FIG. 18 (C), in the processing rectangle, there are cases where the hole position is sparse and dense. There is a need to reduce the speed of the XY stage in the dense position. For example, even in the same processing rectangle, The operation speed of the XY stage is variable. However, it is difficult to determine the operation speed model of the XY stage due to the influence of the speed of the dense position. In order to avoid this situation, in the stage of processing rectangle segmentation, the locations with sparse holes are concentrated as a processing rectangle, and the dense locations are better concentrated as a processing rectangle. Therefore, specifically, for example, the method shown in FIG. 19 is used to evaluate the uniformity of the density of the processed rectangle, and the one with the best vertical and horizontal dimensions is selected. That is, in step 300, the direction along the XY stage is performed to make a cumulative frequency map. The cumulative frequency chart requested here is shown in Figure 20 (C). The horizontal axis is the distance from the end of the processing rectangle toward the XY stage, and the vertical axis is the position of the hole survey along the XY stage. Figure of holes in the rectangle 30- (27) (27) 200302761 number appearing as a percentage. This cumulative frequency diagram is shown in Figure 20 (C) as a monotonic increase, and the more the density becomes the more linear monotonic increase. Therefore, the uniformity of the density can be evaluated based on the number of straight lines. Therefore, a cumulative frequency map is prepared for each processed rectangle in step 300, and an average cumulative frequency map of these is obtained in step 302. Next, in step 304, the linearly increasing offset 値 from the area is evaluated by, for example, area. Then, in step 306, the evaluation value is obtained from both the vertical and horizontal directions, and the best is selected. In addition, as described above, it is also possible to use a sparse and dense situation to form a relief by using the maximum offset instead of the area. Alternatively, the processing in FIG. 16 is performed for the two directions of X and Y. According to the processing time of each processing rectangle calculated in step 224, the sum of the processing time can be selected to be smaller. 4. Regarding the above 1 and 2 to determine the processing rectangle in the horizontal and vertical directions on both sides, and then use the combination of these processing rectangles to divide the processing rectangle into a whole. In this method, as shown in Figure 21, the test in the horizontal direction is preferred. Split the processing rectangles with priority in the vertical direction. After each processing rectangle is determined, select the combination of the processing rectangles in both directions to minimize the movement distance among them. In Figure 21, only the lower left part is processed in the vertical direction. Due to the short operating distance of the rectangle segment, only this part is regarded as a long processing rectangle in the vertical direction. Its -31-(28) (28) 200302761 other parts select the long processing rectangle in the horizontal direction. In addition, instead of using the evaluation criterion as the minimum operating distance, it is also possible to perform evaluation with the same density as described above. In addition, instead of using the evaluation criterion as the minimum operating distance, it may be evaluated as the sum of the processing time of each processing rectangle. 5. More efficient processing rectangle segmentation method that considers the same pattern and distribution density. In this method, the position of the opening in the divided processing rectangle is adjusted by the electric scanner while moving the XY stage. In order to get more efficient processing rectangular segmentation. When the position of the opening position in the divided rectangle is not valid, as shown in Fig. 15 (A), there is a case where the opening position is divided into upper and lower levels, which is different from that of the opening position shown in Fig. 15 (B). Density is not necessarily the case where sparse locations and dense locations are mixed. That is, as shown in FIG. 15 (A), for the case where the XY stage is in the horizontal direction, the opening position is polarized near and near the upper and lower sides of the processing rectangle. For the case where there is no near and near the center of the processing rectangle, Since the amount of movement is large in the longitudinal direction, the processing speed does not increase. In addition, as shown in FIG. 15 (B), the distribution density of the opening positions is not constant. When the direction of the XY stage is sparse, dense, sparse, dense, etc., the speed of the XY stage in the sparse position is dense. The limitation of the speed of movement in the position makes it difficult to determine a better speed model. Therefore, it is better to avoid the polarization or uneven density state of the above-mentioned position of the holes. -32- (29) (29) 200302761 And as can be seen in Figure 22, the distribution of opening positions often shows the same or similar shape (reversal of the same data, etc.), a group of opening positions (called pattern shapes) 11 A is plural and beyond Way of distribution of 1 1 B of Ruoqian remaining points. One of the reasons for the difficulty in processing rectangular segmentation is because of the existence of the above-mentioned number of remaining points 1 1 B. Therefore, this situation is not a direct process of rectangular segmentation in a difficult state. First, the pattern shape 1 1 A is distinguished. Most of the pattern shapes 1 1 A can be easily processed for rectangular segmentation, and then 'processed rectangular segmentation for a number of opening positions 1 1 B can not only be easily handled, but often also become an effective processed rectangular segmentation. Hereinafter, the procedure of dividing the processing rectangle into the hole position will be described with reference to FIG. 23. First, in step 400, a GUI operation or the like is performed in accordance with the content instructed in advance to determine whether or not to make a dense distinction. In the case where the determination result is positive, proceed to step 402, and it can be effectively determined that the processing rectangle is determined for the positions of the openings other than the most sparse group. Next, proceed to step 404, and determine whether to perform the majority / minority distribution of the hole positions according to the contents indicated in advance by GUI operation or the like. If the determination result is positive, proceed to step 406. For the opening positions of the majority group, It can be effectively determined that the processing rectangle is determined. Proceeding to step 408, the processing rectangle is determined for the undetermined opening position of the processing rectangle. -33- (30) (30) 200302761 In step 408, any one of the previously described 1 to 4 processing rectangle division methods is used. The foregoing steps 402 and 406 have similar processes as shown in FIG. 24 (step 402) and FIG. 25 (step 406). That is, (1), grid segmentation and grid recognition (bit allocation) process (step 4 1 0) (2), grid group rectangularization process (steps 414, 415) (3), fine adjustment process of grid group rectangle position ( Steps 4 1 8 and 4 1 9) and (4), and the process of covering the processing rectangle (arrangement of the processing rectangle) by the grid group rectangle (steps 420 and 421) are all performed in this order. Outline each process. Step 4 1 0, 4 11 (1), grid division and grid recognition (bit allocation) process is to divide the processing area on the substrate (that is, the enlarged area of the opening position) into a grid shape, with 0 or 1 (bit Element) to identify the density of the processing positions in each grid (hereinafter referred to as a rectangle parallel to the substrates of the same size, just called grid) or the presence or absence of the process (called the grid where recognition 値 is 1 is a true grid, identify The square where 値 is 0 is a false square). (2) in steps 414 and 415, the grid group rectangularization process is shown in FIG. 26, and a rectangle (referred to as a lattice group rectangle) with the following conditions is used to group the grids with 1 grids. A. There must be 1 (somewhat) in a grid column (four above and below, left and right) corresponding to the outer periphery of a rectangle. -34- (31) (31) 200302761 B. The grid columns (four above and below, left and right) surrounding the outer periphery of the rectangle are all 0 (no dots). Moreover, (3) in steps 4 1 and 4 1 9, the fine adjustment process of the rectangular position of the grid group is (2). The rectangular process of the grid group is distinguished by a grid having a width in the XY direction. The points to be distinguished are not necessarily, so it is a process for more accurately adjusting the arrangement of the grid group rectangles. Steps 420 and 421 (4): The process of covering the rectangular group by processing the rectangular group of the lattice group to obtain the obtained rectangular group of the lattice group is A, the linear arrangement is B; In step 4 1 6, if the size of each rectangle of the grid group is checked, if there is no large grid group rectangle, steps 419 and 421 are not performed. In addition, as shown in FIG. 25, steps 419 and 421 are processes performed only for a large grid of rectangles. This will be described in detail below. The division of the grid in step 4 10 and the identification of the dense grid are specifically performed in the order shown in FIG. 27 as shown in FIG. 28. That is, first, the substrate is divided into a grid shape at step 500, and the number of holes in each grid is counted. Next, in step 504, a degree distribution of the number of holes is prepared, and it is investigated whether there are positions where the degrees are significantly different. In addition, when the grid-like division is performed, the original substrate is regarded as one grid. If the grid is bisected vertically and horizontally, four grids are generated, but the subdivision can be performed by repeating this division. -35- (32) (32) 200302761 In the step 500, it repeats the fine division every time. In step 502, it counts the number of holes in the grid. In step 504, the method of [the number of holes * above * the number of underfilled grids is *] Degree distribution. In step 5 06, the maximum degree in the degree distribution is repeatedly smaller than a certain setting. Next, in step 508, 1 is assigned to the dense grid; 0 is assigned to the sparse grid. Fig. 29 shows how the degree of the opening position is counted for the grid division substrate. Only the dense positions 11C or 11D are shaded by a hatched network. In addition, in the dense position 1 1D surrounded by a circle in FIG. 29, it is determined that the grid belongs to the rectangular grid group having a narrow area in step 416 of FIG. 25, and the processing rectangle is determined as an example of the grid in step 408 of FIG. 23. . Moreover, the segmentation of the grid in step 4 1 and the identification of the perforated grid can be performed in the order shown in FIG. 30 as shown in FIG. 3 1. That is, first, step 606 is divided by a grid of an appropriate size. Substrate. Secondly, in step 6 1 0, even if there is one, 1 is assigned to a grid of points; 0 is assigned to a grid of none. The rectangularization of the lattice group in the foregoing steps 414 and 415 is specifically performed in the order shown in FIG. 32. That is, first, at step 622, a grid group rectangle whose undetermined allocation is 1 is arbitrarily selected, and the grid itself is temporarily made a grid group rectangle. Regarding each of the grid rows or columns (excluding grids corresponding to the four corners of the rectangle) corresponding to the outer periphery (upper, lower, left, and right) of the rectangle of the selected current grid group, if there is a grid at step 624, -36- (33) (33) 200302761 Step 626 enlarges the area of the grid group rectangle in its direction. If there is no up, down, left, and right in step 624, the grid group rectangle is determined in step 628. In step 630, the process returns to step 622 for the case where it is judged that there is a grid with 1 assigned, and the process from step 622 is continued again. End processing. The fine adjustment of the rectangular position of the grid group in steps 4 1 8 and 4 1 9 is performed in the order shown in Figs. 3 3 and 34 respectively. That is, in order to investigate the enlargement of the actual correct opening positions of these rectangles, the opening positions of the outermost holes are investigated for all the rectangles (upper, lower, left, and right) on the outer periphery. That is, in the case of FIG. 34 showing step 419 of the fine adjustment of the rectangular position of the large pattern grid group in detail, if the X coordinate of the opening position of the left grid row is minimum 値 (step 644) X of the opening position of the right grid row The Y coordinate of the opening position of the grid row at the largest coordinate 値 (step 650) is the smallest Y coordinate of the opening position of the grid row at the step 650 (step 656) (step 662), the actual correctness can be obtained. The position of the opening is enlarged. If the coordinate information of the rectangle showing the existence of the opening position is updated by these correct maximum and minimum values, it is possible to identify a regular arrangement. Although it has been described, in the case where the original data file itself has pattern information -37- (34) 200302761, it is not necessary to perform the processing as described above. Further, in the case of FIG. 33 showing step 41 8 of the fine adjustment of the dense grid group rectangle position in detail, there may be holes in the grid rows located outside the outer periphery of the grid rectangle of the grid group determined to be sparsely distributed. Therefore, in steps 640, 646, 6 52, and 658, it is investigated whether there is a hole in the grid column located outside the outer periphery of the rectangular group of the grid group, to obtain the maximum or minimum value of the outer grid column in some cases (step 642, 648, 654, 660), in the absence of the maximum 値 or the minimum 値 of the grid column located on the outer periphery (steps 644, 650, 656, 662). Furthermore, the covering of the processing rectangles (ie, the arrangement of the processing rectangles) by the dense or large grid group rectangles in the foregoing steps 420 and 421 is performed in the order shown in FIG. 35. Here, the correct arrangement of the rules is identified by the set of rectangles of the lattice group, and the set of rectangles covering these rules is covered. The term “regularly correct arrangement” herein refers to a relationship in which rectangles are arranged in a straight line or a relationship in which all are arranged in a linear grid in the XY direction, as shown in FIG. 36 (B). Therefore, the following processing is performed. First, at step 670, in the linear arrangement in the X direction and the Y direction, the coordinates of each of the X direction and the Y direction are classified to group 値 the same. In the example of FIG. 36, X1 {B5, BIO}, X2 {B6, B11}, X3 {B7, B12} in the X direction, Y1 {B1, B2, B3, B4}, Y2 {B5, B6, B7}, Y3 {B10, Bll, B12} are arranged linearly in the Y direction. Next, in the case where there are a plurality of linear arrangements in X and Υ, it is investigated whether or not the lattice arrangement is made by a combination of these. That is, in step 676 -38- (35) (35) 200302761, regarding the above-mentioned grouped rectangles, it is investigated whether the elements of the plurality of groups in the X direction and the elements of the plurality of groups in the Y direction are consistent. In this example, three groups X1 {B5, BIO}, X2 {B6, B11}, X3 {B7, B12} and two groups Y2 {B5, B6, B7}, Y3 {B10, B11, B12} The elements are the same. The practical way to find this consistent combination is as follows. That is, first select the groups in the X direction (the first is X1 {B5, B10}), and investigate whether there are groups with the same elements in the Y direction (Y2 {B5, B6, B7 for B5, and Y3 {B10 for B10, B11, B12Π. Investigate whether the newly-generated elements {B6, B7, B11, B12} in a certain situation have groups X2 and X3 other than XI in the X direction. It is better to have a grid-like configuration if the above operations are completely consistent. It can be obtained from the above linear arrangement as y 1 31, 32, 33, 64}, and the lattice arrangement as X Yl {B5, B6, B7, BIO, Bll, B12}. In the case of arrangement, in the case of linear arrangement in step 680, in step 674, which is a step subsequent to step 680, the processing rectangle is arranged to cover the regularly arranged grid group rectangle. The rectangular group arranged in the covered linear group As shown in Figure 36, pay attention to the following points: A. Basically, as shown in Figure 37 (A), the arrangement of the processing rectangle is determined according to the direction of the straight line extension. B. As shown in Figure 37 (B), the grid group is rectangular. (The distance between the closest positions of the rectangle) If the method of simple division in the horizontal or vertical direction] is described (the size of the processing area -39- (36) (36) 200302761 DM + the distance α about the speed change) is large, arrange a plurality of processing rectangles. C As shown in Figure 37 (C), the width in the direction perpendicular to the straight line is wider than the size DS of the processing area, and if it cannot be covered with one processing rectangle, two or more processing rectangles are arranged. D. Figure 3 As shown in Fig. 7 (D), in the case where a plurality of processing rectangles are arranged, if they are simply arranged, the center position of the covered processing rectangle is smoothly set for the case where the hole position is halved at the end. Specifically, it is not determined by the end It is covered with the width of the processing area in order, but the part which avoids the space | interval of the opening-free position overlaps the center of the processing rectangle of the part where the opening position should be arrange | positioned. In the case of a rectangular group, as shown in Fig. 38, pay attention to the following points. As shown in Fig. 3 8 (A), as in the case of a linear arrangement, for a case where the arrangement interval is larger than D + α, in the X direction and the Y direction All As a result, the rectangular group of the grid group arranged in a grid shape is regarded as having a distance of D + α or less. As shown in FIG. 38 (B), each grid is arranged by covering a plurality of rectangles in the same direction in the X or Υ direction. The choice of either the X direction or the Y direction is determined by, for example, the following evaluation criteria. A. The installation of the XY stage (upper and lower stages) becomes the direction of the upper stage. C. The sum of the action distances (edges in the direction of dividing the rectangle) is smaller than -40- (37) (37) 200302761 (comparison of the sum of Lx and Ly in Figure 38 (B)) or the points in the dividing rectangle The sum of the distances between the existing rectangles is smaller (comparison of the sum of dx and dy). D. Model the operation of the electric scanner and the XY stage, and choose a simple simulation with appropriate operation time. Or as shown in Figure 3 8 (C), if the processing rectangle is simply arranged, the situation where the processing rectangle with the opening position at the end halved is generated to avoid the part without the opening position, and should be arranged. The center of the processing rectangle of the center of the part with the opening position overlaps. Regarding those who cannot cover with the good rectangle in step 421 (B8, B9, and B13 in Figure 36), the rectangle is not determined here. In step 408 of the next Figure 23, the final remaining hole position is used to process the rectangle. Configuration processing. In the case where there are few remaining opening positions, the operation of stopping the processing by the table can be used. That is, in the example of the substrate shown in Fig. 22, the portion 11B surrounded by a circle is a minority group. In the case where such positions are generally arranged in a substrate, each of the minority groups may operate and set the XY stage toward a stop position, and after performing the setting, the irradiator may be effective at a point of processing time. In addition, the table stop processing speed is 0 during processing. When moving between processing areas, for example, a speed model with a constant speed close to the maximum speed. A part of the table stop processing can be implemented in the same control mode. Processing. The process of optimizing the order of access to the opening positions in step 1 and step 2 of Figure 2 and maximizing the speed of the isokinetic operation of the Xγ stage is mainly a combination of the following three basic -41-(38) (38) 200302761 processes. A. Opening position access sequence generation process B. Table operation speed calculation process C. Opening position access sequence improvement process. Process A (opening position access sequence generation process) is a tentative decision to make one side of the XY stage move linearly. , One side can be processed in order to access the opening position (processing order) of the process. The function name of this process in the text below or in the figure, such as a program, is written as initial_sequence (). In addition, the process B (table operation speed calculation process) is a process for accessing a certain hole position. For example, while the XY table is operated at a constant speed, the maximum operation speed of the table operation can be obtained while processing. This process is written as find-speed () in the text below or in the drawing. Moreover, process C (the process of improving the access position of the opening position) is to fix the movement speed of the table at a certain speed, so that the processing rectangle (including the scanning by the electric scanner, (All settings, etc.) The total processing time is shortened, and a process is explored to improve the order of access to the hole position. Reuse the process of this process as continuously as possible < C > (notation here < * > is a symbol for the case where the process * is reused as continuously as possible) is written opt () below or in the drawing. Combining the three processes described above, this process can be performed only by processes A and B as shown in FIG. 39. And as shown in Figure 40, it can also be processed according to process A, B, < C > and B are performed in this order. More preferably, as shown in FIG. 41, with processes A, B, < < C >, B > That is, the processes A and B are repeated as much as possible. "Repeat the opening position -42- (39) (39) 200302761 Improvement of the access sequence and the maximum calculation speed of the table". In addition, as shown in FIG. 42 to be described later, processes A, B, and < {C, B} > (The symbols {*, **} are any symbols that indicate * or **). That is, the processes A and B are repeated as much as possible "any of the improvement of the order of access to the hole position and the calculation of the maximum speed of the table". Alternatively, although some points are different, as shown in FIG. 43, the process A, < C '>, B. Here, the process CT is optimized for a case where the table operation speed is fixed at a certain level and the table operation speed is 0. Therefore, the process CT is not accompanied by an execution possibility determination process (feasibility ()) described later. Specifically, the solution of the traveling salesman problem (in fact, the solution of the local exploration method by minimizing the Hamiltonian path length problem) is cleverly applied. That is, when using a method of the traveling salesman problem such as the Lin-Kernighan method, a restriction that the branch connecting the start point and the end point must be cut off is added to solve the problem. That is, if FIG. 43 is made a little more specific, an initial path is generated as shown in FIG. 44 to solve the problem of the traveling salesman with the restriction that the branches connecting the start point and the end point of the path must be cut. Then, the speed of the XY table can be obtained. deal with. Hereinafter, the processes A, B, and C will be described in detail. In addition, in the following description, only the position of the hole is marked as a point (of course, there is no problem in the description). The process A (the process of sequentially generating the opening position access sequence) is conceived for the case where the χγ stage operates at a constant speed, as shown in FIG. 45. That is, the width h below the size of the processing area is set in the operation direction of the χγ table, and the rectangular shape is divided. For the width h, the opening position accommodated in the width h is divided into -43- (40) (40) 200302761. In the direction perpendicular to the movement direction, if the adjacent widths are classified in ascending order within a certain width, the descending classification is performed in the adjacent widths, which can make the access order of the opening position meandering. The initial path. In addition, as disclosed in [Applied Mathematical Planning Manual] (edited by Kubota et al., Asakura Bookstore, 2002), this method of generating paths is generally called the strip method, but in the present invention, the division is determined as shown below Width h. That is, as shown in Fig. 45, let the length in the direction perpendicular to the operation direction of the rectangular XY stage be a, the length in the operation direction of the XY stage be b, the number of holes in the rectangle be N, and the desired division width be h. Here, from the viewpoint that there is no waste if the amount of movement in the X direction and the amount of movement in the γ direction are equal, the average amount of movement in the direction of XY stage movement and the amount of movement in the direction perpendicular to the direction of XY stage movement are obtained. When the average 値 of the two is consistent, the optimal division width is determined, that is, the division width with the smallest image motion time. For statistical evaluation, it is assumed that the positions of the openings in the rectangle are randomly distributed. First, the average of the movement amount of the XY stage in the direction is the average of the difference between two numbers that occurs randomly in the interval [0, h]. This 値 is considered as the absolute 値 of the two numbers X, y that satisfy 0SXS h, 0 $ y $ h, and the difference between the two numbers is the average 値 of 値 in the area OSx ^ h, O ^ y ^ h. It is calculated by the following formula. -44- (41) (41) 200302761 (1 / h2) HI [y I dxdy = h / 3 ... (2) On the other hand, the average of the amount of movement in the direction perpendicular to the direction of the χγ stage is caused by the width The interval of a is the average interval when the average Nx (h / b) holes are randomly arranged, so it becomes the following formula a / (Nx (h / b)) = ab / (Nh) ... (3) Therefore, both The coincidence system is as shown in FIG. 46 when the following formula is established. h / 3: ab / (Nh) ... (4) If this formula is calculated, it becomes the following formula h =, (3ab / N) = / " (3 / p) ... (5) Here, P Is the density. For the process B (calculation process of table operation speed), the method shown in Fig. 47 is used. First, in step 930, the processing speed calculation process of the non-executable stage described later (this process is written as max_speed () in the text below or in the drawing) is performed to find the infeasible speed V2. Next, in step 9 32, the processing speed calculation process of the workable execution table described later is performed (this process is written below -45 or (42) (42) 200302761 min_speed ()) in the text below, to find out that it can be implemented. Speed vi. In step 934, the number of iterations is initialized by zero. In step 936, the loop condition is put in the case where the number of iterations is ITE times or the case where v2-Vl is 値 below TOL. In the case where it is impossible to get out in step 936, the intermediate 値 v of V2 is found in step 938. In addition, the execution possibility determination process for the speed v described later in step 940 (this process is written as feasibility () below or in the drawing), and it is investigated whether the speed v is feasible. In step 942, it is determined whether or not it is feasible. In the case where it is not feasible, the non-executable speed V2 is updated with v in step 944, and in the case where it is feasible, the feasible speed is updated with v in step 946. Here, description will be made on the unavailable table operation speed calculation process max_speed (), the executable table operation speed calculation process min_speed (), and the execution possibility determination process feasibility (). First, the feasibility () of the execution possibility determination process will be described. As shown in Fig. 48, the possibility of drilling is implemented. As the speed of the XY table increases, it moves from an executable state to an impracticable state. Therefore, the degree of execution possibility is calculated by the following method. That is, as shown in FIG. 49, the timing (Timing) frame limit is set for each point, that is, the earliest start time and the latest end time. Here, the earliest start time refers to the earliest timing (Laming) when laser irradiation for opening can be started, and the latest finish time refers to the latest timing (Timing) for laser irradiation for opening must be completed. Specifically, as shown in FIG. 49, the processing area in the XY stage moves in the γ direction -46- (43) (43) 200302761 at a constant speed on the substrate by the action in the opposite direction of the XY stage, and passes at the upper limit position of the processing area at time 0. The reference point Y = 0. Let the Y coordinate 値 (absolute 値) of point P be YP, then the earliest start time can be Yp / V, and the latest end time can be obtained by (Yp + D) / V. However, D [ideally] indicates the size of the processing area. Here, although the expression [ideally] is used, the 値 of D is actually inconsistent between the above driving mode and the actual mechanical movement. Therefore, to cope with this, the D of D is set to be larger than the size of the processing area. Still smaller. That is, it has a margin of length DF at the earliest start time and a margin of length DB at the latest time (DF, DB is 0 or more) to deal with errors. In this case, the earliest start time can be (Yp + DF) / V, and the latest end time can be obtained (Yp + D-DB) / V. ○ The example is displayed numerically. For example, as shown in Fig. 50, let Yp = 300mm, D = 50mm, and V = 00mm / sec. The earliest start time of the case where there is no margin for error tolerance is 300/100 = 3.0 seconds, and the latest finish The time is (3 00 + 50) / 1 00 = 3.5 seconds. Therefore, the time when the processing of a certain hole position is completed (that is, the irradiation is completed) can be determined to be 3.2 seconds, for example, and 3.7 seconds is not allowed. Now, let the i-th visited point be Pi and the first visited point be P! The irradiation start time is P: the earliest start time. The time at which irradiation at each point is completed can be obtained by adding the time at which the irradiation time is added to the irradiation start time.
Pi + ι的照射開始時刻爲對Pi的照射完了時刻加入由Pi 到+ 1的移動時間（稱此結果的移動完了的時刻爲到達時 刻），若到達時刻比+ 1的最早開始時刻還早的話，令 -47- (44) (44)200302761Pi + ι's irradiation start time is the time when Pi's irradiation is completed. Add the moving time from Pi to +1 (call the time when the result is completed as the arrival time), if the arrival time is earlier than the earliest start time of +1 , Order -47- (44) (44) 200302761
Pw!最早開始時刻爲照射開始時刻，否則藉由令到達時刻 爲照射開始時刻，依次求得。 可以說所有點的照射完了時刻比該點的最遲完了時刻 還早的話可實行，否則不可實行。即實行可能性判定處理 feasibility ()首先如上述依次求出各點的照射開始時刻 ，對所有的點計算（最遲完了時刻-照射完了時刻）。而 且，在這些之中以最小的（在以下記爲f (或fl、f2 )) 作爲判定結果而返回來進行。即正或0的話可實行，負的 話不可實行。而且同時算出最後訪問的點的加工完了時刻 (在以下記爲e (或e!、e2))(因設定開始完了時刻爲 0，故等於總加工時間）。此外，本處理feasibility ()在 製程C (開孔位置訪問順序改善製程）中也使用。 此處，實行加工的兩點間的移動所花的時間首先模型 化XY台停止加工時的X方向、Y方向的各移動時間。而 且，對於XY台以等速移動的情形，以在內部呼叫上述函 數的形，作爲在引數具有移動距離L以及XY台速度V的 函數而可模型化。即如圖51所示，令想求出的移動時間 爲T，在此時間T下移動距離L之中XY台移動VT，電掃 描器在時間T掃描剩餘的距離L-VT的模型。此處，也需 注意若移動方向的正負不同的話，即使移動距離相同掃描 時間也不同。 其次，說明不可實行的台動作速度算出處理 max_speed ()以及可實行的台動作速度算出處理 min_speed ()。這些處理是用以由兩側夾擊真的台動作 -48- (45) (45)200302761 速度而求出的作爲夾擊値的初期値使用（參照圖47以及 圖48 )。初期値因只要令V1爲可實行，v2爲不可實行即 可，故也可令v!爲比〇稍微大的値，V2爲比XY台最大速 度還小的値。但是，爲了高速地求解起見，接近兩者都應 解的値較佳。此處，實行可能以及不可能的XY台動作速 度算出處理有在緊接著製程A (開孔位置訪問順序發生製 程）之後的情形，即實行可能速度未求出的情形，與在緊 接著製程<C> (開孔位置訪問順序重複改善製程）之後的 情形，即實行可能速度求出的情形，的兩種類。 首先，針對緊接著製程A之後的情形來說明。令加工 區域爲在Y方向以速度+v等速遷移於基板上。若考慮A (Xi ' yi ) 、B ( xp yj )的兩點間的移動，位移量爲（xrPw! The earliest start time is the irradiation start time, otherwise it is obtained in order by making the arrival time the irradiation start time. It can be said that the time at which the irradiation of all points is completed is earlier than the time at which the point has been completed, otherwise it is not possible. That is, the feasibility () is executed to determine the irradiation start time of each point in turn as described above, and calculate it for all points (latest completion time-irradiation completion time). In addition, among these, the smallest (hereinafter referred to as f (or fl, f2)) is returned as the determination result. That is, if it is positive or 0, it can be implemented, if it is negative, it cannot be implemented. At the same time, the processing completion time of the last visited point (hereinafter referred to as e (or e !, e2)) is calculated (because the setting completion time is 0, it is equal to the total processing time). In addition, this process feasibility () is also used in process C (opening position access order improvement process). Here, the time taken for the movement between the two points for machining is first modeled as the movement time in the X and Y directions when the XY stage stops machining. Furthermore, in the case where the XY stage moves at a constant speed, the above function is called internally, and it can be modeled as a function having a moving distance L and an XY stage velocity V in the arguments. That is, as shown in FIG. 51, let the desired moving time be T, and the XY stage moves VT among the moving distances L at this time T, and the electric scanner scans the remaining distance L-VT model at time T. Note here that if the positive and negative directions of the movement are different, the scanning time will be different even if the movement distance is the same. Next, an unexecutable table operation speed calculation process max_speed () and an executable table operation speed calculation process min_speed () will be described. These processes are used as the initial stage of the pinch stroke, which is obtained from the speed of the pinch stroke on both sides -48- (45) (45) 200302761 (see Figure 47 and Figure 48). In the initial stage, as long as V1 is executable and v2 is not executable, v! May be slightly larger than 0, and V2 may be smaller than the maximum speed of XY stage. However, for high-speed solutions, it is better to approach both solutions. Here, the possible and impossible XY stage operation speed calculation processing may be performed immediately after the process A (opening position access sequence generation process), that is, when the possible speed is not calculated, and when the process is immediately ; C > There are two types of situations after (opening hole position access sequence is repeated to improve the process), that is, cases where the possible speed is obtained. First, the situation immediately after the process A will be described. Let the processing area move on the substrate at a constant speed in the Y direction at a speed + v. If we consider the movement between two points of A (Xi 'yi) and B (xp yj), the displacement is (xr
Xi、yj - y 1 )二 （Xdif、y dif ) o 在t時間於兩點間動作的情形電掃描器的掃描量由A 到B的動作的情形爲（Xdu、ydi卜Vt)，由B到A的動作 的情形爲（-Xdu、-ydif-Vt )，而這些動作的掃描時間成爲 必定比無Y方向的掃描的情形的動作量（Xcnf、〇 )的情形 的掃描時間還大（或相等）的値。因此，求不可實行的初 期速度V2的函數max_speed()如以下而求出。 如圖49令加工區域在Y方向於基板上等速遷移，對 開孔位置訪問順序S上的兩點間的移動量（Xdif、ydu )， 計算移動量（Xdif、〇)的情形的掃描時間。而且，令各個 量的總和爲r。 其次，藉由在掃描區域的基板上的XY台進行方向的 -49- (46) (46)200302761 移動距離除以τ求出XY台的動作速度V2。 如此，可求出具有不可實行的保證的接近可實行的界 限速度。 另一方面，求可實行的初期速度Vi，一倂更新不可實 行的初期速度 V2的値的處理min_speed ()並非如函數 max_speed ()般直接求出，而是如圖52所示求出。即關 於在步驟800對當作函數max_speed ()的返回値而得到 的不可實行的初期速度v2，在步驟802除以2的値，在步 驟8 04檢查實行可能性，若可實行的話完成處理。另一方 面，若不可實行的話在步驟808以V!更新V2。重複上述 順序直到可實行爲止。此處，需注意 v2的値也有由 max_speed ()求出的値改變的可能性。 其次，針對緊接著製程<C>之後的情形使用圖53來說 明。此情形在實行製程<C>的前階段中，可實行側的台動 作速度v!已經確定。因此，min」peed ()係取得在實行 製程<C〉的前階段的値，僅設定成V1的話佳。而且，由於 不可實行側的動作速度應不會距可實行側的動作速度太遠 ，故到實行feasibility ()時使負値返回爲止，重複比1 稍大的係數乘以v!的處理的話佳。此時，需注意Vl的値 也有由min_speed ()求出的値改變的可能性。 此外，在圖42的步驟710或圖47的步驟93 8求出新 的速度v時除了二分法外，例如使用如下式的夾撃法也可 以。 -50- (47) (47)200302761 v= ( -fl x V2 + f2 x vi ) / ( f2-fl ) · ( 6) 此夾擊法如圖54所舉例說明的，在v-f平面中求出 v= vi時的f値（使用實行可能性函數feasibility ()算出 的値）f 1，與以直線連結v= v2時的f値f2時的與f=0 的交點的v値，成爲比例値而非二等分。 針對製程<C> (開孔位置訪問順序改善製程），可適 用用以解決所謂的『附時間框非對稱哈密頓路徑長最小化 問題（Asymmetric Hamiltonian Path Problem with Time Window)』（在以下使用略語，記爲AHPP-TW)的解法 。此處，非對稱性爲兩個開孔位置間的掃描所花的時間（ 移動成本）對應依照視兩點之中的任一點爲始點而不同， 時間框係對應已經說明的最早開始時刻、最遲完了時刻的 限制。而且，開孔位置的訪問爲各位置各一度，不返回到 原來的位置（不巡迴）的路徑，這種路徑一般稱爲哈密頓 路（或哈密頓路徑）。 此處，移動成本爲非對稱的理由乃因台速度存在（參 照圖51)，惟實際上台速度比電掃描器掃描速度還緩慢 (圖5 1爲了使非對稱容易了解起見而變形）。因此，需 注意適用捕捉附時間框對稱哈密頓路徑長最小化問題（略 記爲HPP-TW )的解法而非非對稱，可謀求主要以移動成 本算出時間爲中心，最佳化計算時間的高速化。此處，藉 由看成對稱成本雖然產生若干的誤差，但該誤差若使用已 經敘述的時間框設定中的誤差裕度附加處理來處理的話佳 -51 - (48) (48)200302761 因此，製程C適用藉由局部探索法（或者也稱爲近傍 探索法）的改善方法。即製程C主要由以下的三個處理構 成。 a、 改善解探索處理 b、 改善判定處理 c、 解更新處理 處理a (改善解探索處理）是對暫定的開孔位置訪問 順序（令成S )，由S的近傍N ( S )選擇比S還有效的 開孔位置訪問順序的處理（本處理在本文以下或圖面中寫 成choose ())。此處，S的近傍N ( S )具體上是指一 般稱爲近傍操作的例如藉由將第X個訪問的開孔位置移到 第y個（延緩或提前）等的操作而實現的其他訪問順序的 集合。 此處，適用AHPP-TW的較佳近傍操作有如以下者。 甲、如圖55所示的2-exchange (交換）近傍。 乙、如圖56(前方（forward))或如圖57 (後方（ backward))所示的 Or-opt 近傍。 丙、如圖58 (無反轉）或如圖59 (有反轉）所示的 插入近傍。 除此之外各種近傍也被考慮’惟因依照近傍的擴大計 算時間增大，故使用上述三個近傍（第三個（丙）到插入 的節點數數個左右）爲止爲現實。特別是〇r-opt法無路 徑的反轉故適合非對稱的問題’而且’近傍操作法則是到 -52- (49) (49)200302761 點的路徑上其他位置的移動，是否遵守時間框限制的判定 容易，故適合於具有非對稱且時間框的情形。而且’在圖 3所示的XY台動作最小化用的矩形分割中，雖然被分割 的矩形重複的情形被考慮，但對於這種情形兩條以上的路 徑的同時更新也可以。此情形的近傍操作例如有如以下者 〇 甲、如圖60所示的交叉交換近傍 如圖61所示的2-opt*近傍 這些以外也考慮各種近傍，考慮計算時間與計算結果 的兩方，可使用適切的近傍操作。 處理b (改善判定處裡）是對暫定的開孔位置訪問順 序（以S)判定由近傍N ( S)選擇的解V是否爲更有效 的解的處理。本處理適用已經敘述的feasibility ()。即 計算實行可能性（f )以及總加工時間（e )。此處，改善 解用的條件爲可實行且總加工時間被縮短。視情況有先僅 看實行可能性來判斷的情形（參照圖42 )或看兩方的成 立來判斷的情形（參照圖62 )。 處理c (解更新處裡）是以改善解S ’更新暫定的開孔 位置訪問順序S的處理。與訪問順序一起各點的加工開始 時刻也被更新。 在以上敘述各製程A、B、C的詳細，惟透過步驟112 的製程在以下使用圖42說明進行製程A、B、<丨(：、B}>的 情形的製程的具體例。 首先，在步驟700使加工位置的訪問順序發生一個（ -53- (50) (50)200302761 initial_sequence ( ) ) 0 其次，在步驟702中對點訪問順序的初期解S求一個 不可實行的高速側的XY台速度（ma?c_speed ( ) ) v2。 其次，在步驟704中同樣地對點訪問順序的初期解S 求一個可實行的低速側的XY台速度（min__speed ( ) ) η 。此順序的目的爲求可實行的界限速度，惟在本實施形態 藉由逐次更新可實行的速度値v!與不可實行的速度値V2 以充分接近接近値來實現。 因此，進行脫出條件爲最大反覆次數比ITE次還大， 或不可實行速度與可實行速度的速度差（V2- )的値充 分小的情形的第一環處理。 具體上，在步驟706中令反覆次數i的初期値爲0， 在步驟308中判定是否反覆次數i比最大反覆次數ITE還 大，或高速側台速度V2與低速側台速度V!的差（V2- Vi ) 比臨界値TOL還小的任一方成立。當判定結果爲否時進 行到步驟710，當作實行可能性（feasibility)的判定對象 ，設定V!與v2的中間値v (二分法）。而且，在步驟712 中求依照當時的速度v以及訪問順序S的實行可能性（最 遲完了時刻與照射完了時刻的差的最小値）Π以及總加工 時間（最後訪問的點的照射完了時刻）el，在步驟714判 定是否可實行。 不可實行的情形進行令脫出條件爲可發現可實行的訪 問順序的情形，或得知可實行的訪問順序在近傍無的情形 的任一方的無限環的第二環處理。 -54- (51) 200302761 具體上對於在步驟712求出的實行可能性fl爲負， 在步驟7 1 4判定爲不可實行的情形，在步驟7 1 6進入無限 環的第二環的處理，在步驟7 1 8藉由對訪問順序S加入近 傍操作（例如順序交換）求導入的訪問順序 V ( choose ( ))，在步驟720對其近傍內的訪問順序 V，與fl、el 一樣求實行可能性函數f2以及總加工時間e2的値。 其次，在步驟722中判定f2是否爲負。對於判斷判 定結果爲正，f2爲負不可實行的情形，進行到步驟724， 判定是否e2比e 1大。對於判斷判定結果爲正的情形，進 行到步驟726，以判斷爲更佳的訪問順序的s’更新S。 對於在步驟726終了後或步驟724的判定結果爲否的 情形’進行到步驟728，由近傍N ( S )刪除訪問順序S， 。其次’進行到步驟730，在判斷爲近傍內具有改善解的 可能性的訪問順序消失時，進行到步驟7 3 2，以當時的速 度v更新不可實行的速度V2，脫出第二環進行到步驟752 〇 在步驟732終了後或步驟73〇的判定結果爲否，自第 二環脫出。 另一方面’在判斷爲步驟722的判定結果爲否，訪問 順序S ’與速度v爲可實行時，進行到步驟740，以S，更新 訪問順序S ’以v更新可實行的界限速度νι，自第二環脫 出。 而且’在判斷爲前述步驟7 1 4的判定結果爲否，訪問 順序S與速度v爲可實行時，進行到步驟7 5 〇，以v更新 -55- (52) (52)200302761 可貫fT的界限速度V1。 步驟732、步驟740或步驟750終了後進行到步驟 752，對I增量返回到步驟708，再度實行第一環。 本例係固定台速度於不可實行的速度，也是到可實行 爲止重複實行製程C的情形的處理的例子。 而且，透過步驟112的製程在以下使用圖62說明進 行製程A、B、<<C>、：6>的情形的製程的具體例。 本例係固定台速度於可實行的速度，也是只要有改善 解就重複實行製程C的情形的處理的例子。 首先，在步驟2002使加工位置的訪問順序S發生（ 處理 initial_sequence ( ) ) ° 其次，在步驟2004對訪問順序S算出可加工實行的 台等速最大速度v (處理find —speed ())。 再者，在步驟2006、2007、2008對從步驟2010的環 處理，求總加工時間e (處理feasibility ())，將暫定 的最大速度v拷貝到oldv，將總加工時間e拷貝到oldE 以及oldoldE，初期化反覆次數i爲〇。此外，〇ldE在步驟 2016的各次的近傍探索中的改善解判定，〇ldoldE是在步 驟2024中的環脫出判定使用。 由步驟20 1 0開始並行最佳化台速度以及訪問順序用 的環處理。環的脫出（等於製程的終了）是在到達充分的 環反覆次數時（步驟20 1 0 )、環中總加工時間完全不變 時（步驟2024 )以及在環前以及環後速度的改善太小（ 決定設定値T0L )時（步驟2028 )。此外，這些脫出判 -56- (53) (53)200302761 定即使不使用全部也無妨，最低若使用一個的話佳。 首先，在環開始時點（步驟2012 )由S的近傍N ( S )內選擇一個具有改善解的可能性的解S’（處理choose ( ))。近傍使用例如已經顯示於圖54〜59者。 其次，對所選擇的V使用處理feasibility ()，算出 S 1勺實行可能性f以及總加工時間e (步驟20 1 4 )。 在步驟2016中，判定S’是否爲改善解。改善解的條 件爲可實行（fg〇)且總加工時間被縮短（〇ldE>e)。 判定爲改善解的情形（步驟20 1 8 )，以 V更新解（ 暫定的最佳訪問順序），爲了下一次的判定，將總加工時 間e拷貝成oldE。而且，返回到步驟2012再度探索改善 解。 判定爲非改善解的情形（步驟2020 )由近傍去除Y ，在在步驟2022調查在近傍是否還有改善解探索的餘地 。針對所有的近傍解處理已經終了，無近傍探索餘地的情 形，在步驟2024判定在由步驟2010開始的環中是否改善 一次也未被進行。判定條件例如看環處理實行前的總加工 時間oldoldE是否與oldE相等。 在步驟2024中判定改善被實行一次（即oldoldE# oldE )的情形進行到步驟2026，算出實行可能台最大速度 v。其次在步驟2028中，台最大速度v對環處理實行前的 台最大速度oldv可判斷爲幾乎不被改善（對設定値T0L ，v-oldV<TOL)的情形，由環處理脫出完成製程。改善爲 充分的情形進行到步驟2030，爲了下一個環，求總加工 -57- (54) (54)200302761 時間e (處理feasibility ())，將暫定的最大速度v拷 貝到oldv，將總加工時間e拷貝到oldE以及oldoldE (步 驟203 2 )，對環反覆次數I增量一次（步驟2034 )，返 回到步驟2010再開環。 另一方面，在步驟2024中判定改善一次也未實行（ 即oldoldE = oldE)的情形完成製程。 而且，藉由對在圖2的步驟1 1 2決定的開孔位置的在 步驟1 14的開孔位置的分布密度平順地使動作速度變化的 XY台加減速模型最佳化（速度模型決定製程）可如以下 來進行。 現在在某加工矩形內，在某可實行的開孔位置訪問順 序以及XY台（等速）動作速度的電掃描器、加工區域的 基板上的遷移是如圖49的實例的圖63所示。在此例雖然 成爲如圖所示的開孔位置的散佈狀態爲密的場所與疏的場 所混合存在的加工矩形，惟在疏的位置藉由電掃描器的雷 射束的照射位置在加工區域的最早開始時刻側的端停止， 成爲等待XY台的到達的狀態。這種開孔位置散佈狀況的 情形並非使XY台的動作速度在矩形內等速，而是高速化 疏的位置較佳。 具體上如圖64所示，將XY台的動作速度區分成兩 個。即對在上述求出的開孔位置訪問順序與動作速度，區 分開孔位置訪問順序的連續的部分（程序）之中密的程序 ，對密的程序以先求出的動作速度對應，對其他疏的位置 以高速的動作速度對應。 -58- (55) 200302761 在區分密的程序時，例如如圖6 5選擇程序的 及末尾。 即首先在步驟1000中前頭的候補選取等待泽 與（最早開始時刻-到達時刻）較大者）由正的狀;i 的狀態的最初點。而且’由前頭依次看程序’ 1 002出現等待時間正的點的情形不識別密的程序 驟1 006選擇下一個前頭的候補。在步驟1008判斷 時完成。而且，在步驟1〇〇4在（最遲完了時刻-照 時刻）不比臨界値還小的値之下判定爲到達最後的 成。 另一方面，對於步驟1 002、1004都爲否的情 遲完了時刻-照射完了時刻），因有比某臨界値還 的點，故在步驟1010在（最遲完了時刻-照射完了 比某臨界値還小的開孔位置之中，令最小者爲程序 〇 如此，如圖64所示可識別密的程序。 而且，最大化疏的位置的動作速度，例如以比 置還高速的一定的速度使疏的位置動作是如圖66 進行。即以如圖65所示的方法區分的密的程序以 疏的程序，惟需要變更速度用的加減速時間。因此 疏的程序的每一個針對步驟11 2藉由已經說明的方 大化動作速度（同時最佳化開孔位置訪問順序也無 僅使用find_speed ()決定速度也可以）。 此外，關於疏的程序的最大速度算出，若藉由 前頭以 ，（0 移到0 在步驟 ，在步 無候補 射完了 點時完 形（最 小的値 時刻） 的末尾 密的位 所示而 外變成 ，對於 法來最 妨，或 加減速 -59- (56) (56)200302761 時間也包含於疏的程序，在加減速時間也考慮爲等速（高 速）動作的情形可實行的話，因加減速時間的動作速度比 疏的程序的等速動作速度還低速，程序全體爲可實行，故 問題不發生。 對於上述疏的程序可更高速化的情形如圖67 ( B )所 示，可進行多重的速度對應。即對上述疏的程序再度適用 步驟1 1 4的手法，區分疏的程序與密的程序，可更高速化 疏的程序。 此外，開孔位置訪問順序以及台動作速度最佳化製程 (步驟11 0 )的結果，全點的照射完了時刻比最遲完了時 刻還早某程度的情形由圖49得知爲在加工區域下限位置 具有餘裕。此情形爲了在加工區域的中央部分進行加工， 可在全部點的照射完了時刻比最遲完了時刻還早的範圍內 偏移所有的開孔位置的加工時刻同一時間份。 前述步驟120是以由步驟100決定的加工矩形的位置 以及由步驟11 0決定的加工矩形的入出口的動作速度爲基 礎，如圖67所舉例說明地最佳化XY台的動作路徑。 此最佳化的目的是最小化使移動某（存在開孔位置） 加工矩形間，即電掃描器與雷射束休止的時間的總和。 參照圖6 8，說明由某加工矩形到其他某加工矩形的 XY台的動作。關於加工矩形A、B，XY台動作方向以兩 方向箭頭表示，而在加工矩形的左右端的位置是成爲朝加 工區域遷移路徑上的加工矩形的入口以及出口。在加工矩 形A的入出口 Al、A2的XY台的速度是藉由步驟no的 -60- (57) (57)200302761 最佳化決定，惟關於XY台動作方向（由右到左或由左到 右）則未被決定。若考慮時間以及空間的對稱性（進行完 全相反的動作），可使在步驟110 —時地使用的動作方向 相反，速度模型以及開孔位置訪問順序也可完全相反。因 此，ΧΥ台動作方向也成爲最佳化的對象。 兩點間的ΧΥ台的動作時間與電掃描器的情形相同， 模型化、函數化。函數的引數由座標（X!、Υ!)、速度（ Vx、Vy )的狀態到座標（χ2、γ2 )、速度（Vx、Vy )的狀 態的位移的情形具有8個引數（X!、Y!、χ2、γ2、Vx、Vy 、Vx、Vy )。 令此問題如以下當作附限制巡迴售貨員問題（正確爲 因無須返回到原來，故爲哈密頓路徑長最小化問題）來捕 捉而解決。 具體上以各加工矩形爲一*端爲入口 一^端爲出口的兩個 都市來考慮。而且，連結都市間的線（稱爲枝）之中如圖 69所示加入接鄰加工矩形的入出口的枝必須去除的限制 。即獨自列舉這種枝。 解的構成法可舉出局部探索法等的探試（heuristic ) 。考慮計算時間與計算結果的兩方，可使用適切的方法。 此外，在前述實施形態中不僅加工矩形的訪問順序， 連加工矩形的入出口也同時需要最佳決定，故成爲複雜的 構成，惟也能串聯地進行加工矩形的訪問順序最佳化與加 工矩形的入出口（ XY台的進行方向）最佳決定。 例如對於在最佳化加工矩形的訪問順序後，決定加工 -61 - (58) (58)200302761 矩形的入口的情形，可如圖70所示來進行。 即首先在步驟1 1 〇〇最佳化加工矩形的訪問順序。具 體上以加工矩形的中心位置爲一個都市來捕捉’使用藉由 哈密頓路徑長最小化問題的局部探索的解法’最佳化加工 矩形的訪問順序。 其次，決定加工矩形的入出口。具體上在步驟1102 例如如左側或下側等決定最初訪問的加工矩形的入出口。 而且，令該加工矩形爲現在的加工矩形。 接著在步驟1104令下一個加工矩形的兩個入出口候 補之中距現在的加工矩形的出口近者爲入口，其他爲出口 。在步驟11 06到最後的訪問順序的加工矩形的入出口決 定爲止重複步驟1104。 或者，令步驟1104如以下也可以。即令下下一個加 工矩形的兩入出口候補之中距現在的加工矩形的出口近者 爲暫時的入口（依次暫時入口），其他爲暫時的出口。關 於下一個入口候補的兩個計算下一個加工矩形的兩入出口 候補之中來自現在的出口的到下一個入口的成本，與從下 一個出口到依次暫時入口的成本，令成本小者爲下一個入 □。 或者若令加工矩形的數目爲K則解的組合數爲2K。 若加工矩形的數目不多的話，也能完全列舉所有的可能性 。或者即使加工矩形的數目多使用局部探索法等求佳的解 也可以。 而且，在決定加工矩形的入出口後最佳化加工矩形的 -62- (59) (59)200302761 訪問順序時，如圖7 1所示地進行。即首先在步驟1 200決 定所有的加工矩形的入出口。具體上考慮 甲、若進行方向爲左右方向的話令左，爲上下方向的 話令下 乙、對各加工矩形隨機地設定。等 其次，在步驟1 202最佳化加工矩形的訪問順序。具 體上加工矩形一個一個當作都市來捕捉。而且，使用藉由 非對稱哈密頓路徑長最小化問題的局部探索的解法，最佳 化加工矩形的訪問順序。此處非對稱乃因由加工矩形A ( 的出口）到加工矩形B (的入口）的移動，與由加工矩形 B (的出口）到加工矩形A (的入口）的移動爲加工矩形 的入出口已經決定，故不同。 或者，對於加工矩形的進行方向均在X方向或Y方 向一致的情形可如圖72 (所有的加工矩形的進行方向爲X 方向的情形）所示來進行。 即令最初的加工矩形爲γ座標最小且同一 γ座標的 有複數個的情形在其中爲X座標最小者。設定爲最初的加 工矩形的加工區域的基板上的遷移方向+(正）的方向。 而且，如以下由Y座標小者依次排列。此時，同一 γ 座標者令加工區域移動方向也是左右同一。即取由最左者 朝右方向，或其左右相反。另一方面，Y座標不同的情形 在基板上加工區域遷移現在的加工矩形到右的情形，若下 一個加工矩形的左端比現在的加工矩形的右端還左的話， 不左右改變遷移方向，否則使遷移方向逆轉。或者採用其 -63- (60) (60)200302761 左右相反。 如此，加工矩形由Y座標小者朝大者，使基板上的加 工區域的移動路徑成爲蛇行狀而無浪費地連接。 此外，在前述實施形態中如圖2所示，首先進行開孔 位置的加工矩形分割，其次，進行各加工矩形內的開孔位 置訪問順序最佳化以及XY台動作速度最佳化，最後進行 XY台動作路徑的最佳化，惟如圖73所示的第二實施形態 ，首先在步驟1 300最佳化基板全體的開孔位置訪問順序 ，其次在步驟1301也能有效地設定XY台控制目標位置 (控制台動作時的指定位置），決定XY台動作模型。 在前述步驟1 300的基板全體的開孔位置訪問順序最 佳化時，例如將問題當作兩點AB間的成本由A移動的情 形與由B移動的情形相等的對稱巡迴售貨員問題來捕捉， 例如使用L i η - K e r n i g h a η的算法等的已知的手法，可進行 最佳化。 另一方面，針對步驟1310的ΧΥ台動作模型決定， ΧΥ台的速度或加速度計劃因台不限於直線動作，故非常 難。因此，對ΧΥ台決定目標位置，用以在朝該目標値進 行之間，進行藉由電掃描器的掃描以及藉由雷射照射的加 工。在此方法中，爲了保證可實行的動作，如圖74所示 設定ΧΥ台目標位置。 首先考慮以與加工區域同尺寸的矩形群區分開孔位置 。對藉由前述步驟1 300決定的開孔位置（對訪問順附加 Ρ:、Ρ2·..的順序），在步驟1312令訪問順序的前頭Ρ!爲 -64- (61) (61)200302761 著眼點（i=l )。而且，以{D!、D2...丨表示群，設定下標k 爲1 〇 而且，如步驟13 14〜1 320所示，藉由反覆下一處理到 到達最後的開孔位置爲止，可藉由與加工區域同尺寸的矩 形群區分成開孔位置。 即在步驟1314關於j ( S 1 )，尋找像均收容 於與加工區域同尺寸的矩形的最大的j，在步驟1 3 1 6令 {Pi...Pj}爲一個群 Dk。 其次，在步驟13 1 8令下一個注目點i爲j +1，使k增 量。 其次，在步驟1 3 22如以下設定XY台的目標位置。 即對被群區分成Dk的點集合{Pi…Pd，令{Pi ._.Pd的擴大的 中心位置爲目標位置。此處，擴大的中心是指令{ Pi…Pj} 之中的X方向、Y方向的{最小値、最大値}分別爲UnHn、Xi, yj-y 1) Two (Xdif, y dif) o When t operates between two points at time t The scanning amount of the electric scanner is from A to B (Xdu, ydi, Vt), and B The action to A is (-Xdu, -ydif-Vt), and the scan time of these actions must be longer than the scan time of the action amount (Xcnf, 0) in the case of scanning without the Y direction (or Equal) Therefore, the function max_speed () for obtaining the impracticable initial speed V2 is obtained as follows. As shown in Fig. 49, the processing area is moved at a constant speed on the substrate in the Y direction, and the scanning time is calculated for the movement amount (Xdif, ydu) between the two points on the access sequence S of the hole position. Moreover, let the sum of the respective quantities be r. Next, the moving speed V2 of the XY stage is obtained by dividing the moving distance of -49- (46) (46) 200302761 in the direction of the XY stage on the substrate in the scanning area. In this way, it is possible to obtain an approachable limit speed with an unenforceable guarantee. On the other hand, when the feasible initial speed Vi is obtained, the processing min_speed () of updating the non-implementable initial speed V2 is not directly obtained like the function max_speed (), but is obtained as shown in FIG. 52. That is, regarding the infeasible initial speed v2 obtained by returning 値 as a function max_speed () in step 800, divide 除 by 2 in step 802, check the possibility of implementation in step 804, and complete the process if feasible. On the other hand, if it is not possible, V2 is updated with V! In step 808. Repeat the above sequence until it becomes feasible. Here, it should be noted that the 値 of v2 may also change the 値 determined by max_speed (). Next, the situation immediately after the process < C > will be described using FIG. 53. In this case, in the previous stage of implementing the process < C >, the stage operation speed v! Of the implementable side has been determined. Therefore, “min” peed () is the value obtained at the previous stage of the implementation of the process <C>, and it is better to set it to V1 only. In addition, since the operation speed of the non-executable side should not be too far from the operation speed of the executable side, it is better to repeat the process of multiplying the coefficient slightly larger than 1 by v! Until the negative 値 is returned when feasibility () is implemented. . At this time, it should be noted that the 値 of Vl may also change the 値 obtained by min_speed (). In addition, when the new speed v is obtained in step 710 in FIG. 42 or step 938 in FIG. 47, in addition to the dichotomy method, for example, a sandwich method as shown below may be used. -50- (47) (47) 200302761 v = (-fl x V2 + f2 x vi) / (f2-fl) · (6) This pinch method is illustrated in Figure 54. f 値 at vi = (値 calculated using the implementability function feasibility ()) f 1 is proportional to v 値 at the intersection with f = 0 at f 时 f2 when v = v2 when f = f2. Not bisected. For the process < C > (opening position access order improvement process), it is applicable to solve the so-called "Asymmetric Hamiltonian Path Problem with Time Window" (used below) Abbreviated as AHPP-TW). Here, the asymmetry means that the time (moving cost) spent scanning between the two opening positions varies depending on whether any one of the two points is the starting point, and the time frame corresponds to the earliest start time already described, The time limit has been completed at the latest. In addition, the access to the hole location is one degree for each location, and the route that does not return to the original location (not touring) is generally called the Hamilton Road (or Hamilton Path). Here, the reason why the moving cost is asymmetric is due to the existence of the table speed (refer to Fig. 51), but the table speed is actually slower than the scanning speed of the electric scanner (Fig. 51 is deformed to make the asymmetry easier to understand). Therefore, it should be noted that the solution to the problem of minimizing the symmetric Hamiltonian path length with a time frame (abbreviated as HPP-TW) is not asymmetric, and it is possible to optimize the speed of the calculation time by focusing on the calculation time of the moving cost. . Here, although it is regarded as a symmetric cost, although some errors are generated, it is better if the errors are processed using the error margin additional processing in the time frame setting already described. -51-(48) (48) 200302761 Therefore, the manufacturing process C is an improvement method using a local search method (also called a near-field search method). That is, the process C is mainly composed of the following three processes. a. Improved solution exploration processing b. Improved judgment processing c. Solution update processing a (Improved solution exploration processing) is a tentative order of opening positions (set to S), and the ratio of S to S is selected by S near N (S) Also effective processing of the order of access to the opening position (this processing is written as choose () below or in the drawing). Here, the proximity N (S) of S specifically refers to other accesses generally referred to as proximity operations, for example, by moving the opening position of the Xth access to the yth (delayed or advanced) operation. Ordered collection. Here, the preferred near-field operation for applying AHPP-TW is as follows. A. 2-exchange (exchange) as shown in Figure 55. B. Or-opt near as shown in Figure 56 (forward) or Figure 57 (backward). C. Insert as shown in Figure 58 (without reversal) or Figure 59 (with reversal). In addition, various proximity are also considered. However, since the calculation time increases according to the expansion of the proximity, it is realistic to use the above three proximity (the third (c) to the number of inserted nodes). In particular, the zero-opt method has no path reversal, so it is suitable for the problem of asymmetry. And, the near operation rule is -52- (49) (49) 200302761. The movement of other positions on the path, does it comply with the time frame limit? The determination of is easy, so it is suitable for the situation with asymmetric and time frame. Furthermore, in the rectangular division for minimizing the operation of the XY stage shown in FIG. 3, although the case where the divided rectangles are repeated is considered, it is possible to update two or more paths simultaneously in this case. Proximity operations in this case are, for example, the following: 0. Cross-exchange proximities as shown in Figure 60. 2-opt * Proximity as shown in Figure 61. Various proximities are also considered. Considering both the calculation time and the calculation result, you can Use appropriate nearby operations. The process b (in the improvement judgment section) is a process of determining a tentative opening position access sequence (in S) to determine whether the solution V selected by the nearby N (S) is a more effective solution. This process applies to the already described ease (). That is, the execution probability (f) and the total processing time (e) are calculated. Here, the conditions for improving the resolution are feasible and the total processing time is shortened. Depending on the situation, the situation may be judged only by the possibility of implementation (see Figure 42) or the establishment of the two parties (see Figure 62). The process c (in the solution update section) is a process for improving the solution S 'to update the tentative opening position access sequence S. The processing start time of each point along with the access sequence is also updated. The details of each of the processes A, B, and C are described above, but a specific example of the process in the case where the processes A, B, < 丨 (:, B) > are performed will be described below using FIG. 42 through the process of step 112. First, In step 700, an access sequence of the machining position occurs (-53- (50) (50) 200302761 initial_sequence ()) 0 Next, in step 702, the initial solution S of the point access sequence is found to be an impractical high-speed side. XY stage speed (ma? C_speed ()) v2. Next, in step 704, the initial solution S of the point access sequence is similarly determined. An XY stage speed (min__speed ()) η that can be implemented on the low-speed side is obtained. In order to obtain the feasible limit speed, in this embodiment, the feasible speed 可 v! And the non-executable speed 値 V2 are realized by updating the approachable speed 充分 V2 sufficiently close to 値. Therefore, the withdrawal condition is the maximum repetition rate ratio. ITE times are large, or the first loop of the case where the speed difference (V2-) between the infeasible speed and the feasible speed is sufficiently small. Specifically, in step 706, the initial value of the number of iterations i is 0, and It is determined in step 308 The number of repetitions i is greater than the maximum number of repetitions ITE, or the difference (V2- Vi) between the high-speed side table speed V2 and the low-speed side table speed V! Is smaller than the threshold 値 TOL. When the determination result is negative, it proceeds to In step 710, as a determination object of the implementation feasibility, an intermediate 値 v (dichotomy) between V! And v2 is set. Further, in step 712, the implementation possibility according to the current speed v and the access sequence S is determined ( The minimum difference between the latest completion time and the irradiation completion time 値) and the total processing time (irradiation completion time of the last visited point) e1, it is determined whether it is feasible in step 714. If it is not feasible, the release condition is allowed A case where an executable access order is found, or a second ring process of an infinite loop of either of the cases where the accessible access order is not near. -54- (51) 200302761 Specifically, for the The execution possibility fl is negative. If it is judged as impossible in step 7 1 4, the processing of the second loop of the infinite loop is entered in step 7 1 6, and the near operation is added to the access sequence S in step 7 1 8 (for example, Order exchange) to find the imported access sequence V (choose ()), and in step 720 to the access sequence V in its immediate vicinity, as in fl, el, find the implementation of the probability function f2 and the total processing time e2. Second, in step In 722, it is determined whether f2 is negative. For the case where the determination result is positive and f2 is negative and cannot be implemented, proceed to step 724 to determine whether e2 is greater than e1. For the case where the determination result is positive, proceed to step 726, Update S with s' judged to be a better access sequence. After the end of step 726 or the result of the determination at step 724 is no, it proceeds to step 728, and the access sequence S, is deleted from the neighborhood N (S). Next, proceed to step 730, when it is judged that the access sequence that has the possibility of improving the solution in the immediate vicinity disappears, proceed to step 7 3 2 and update the infeasible speed V2 at the current speed v, and go out of the second loop to In step 752, after the end of step 732 or the determination result in step 73 is no, it comes out of the second loop. On the other hand, when it is determined that the determination result of step 722 is No, the access sequence S and the speed v are feasible, proceed to step 740, and update the access sequence S with S, and update the feasible limit speed ν, Come out from the second ring. Furthermore, when it is judged that the determination result of the foregoing step 7 1 4 is no, and the access sequence S and speed v are feasible, proceed to step 7 5 0, and update with v-55- (52) (52) 200302761 through fT The limit speed V1. After the end of step 732, step 740 or step 750, proceed to step 752, return the increment of I to step 708, and perform the first loop again. This example is an example of a process in which the process C is repeatedly performed until the execution speed is fixed, and the execution speed is repeated. A specific example of a process in the case where processes A, B, < < C >,: 6 > This example is an example of a case where the fixed table speed is a feasible speed, and the process C is repeated as long as there is an improvement. First, in step 2002, the access sequence S of the processing position is generated (processing initial_sequence ()). Second, in step 2004, the maximum access speed S that can be processed is calculated for the access sequence S (process find-speed ()). Furthermore, for the loop processing from step 2010 in steps 2006, 2007, and 2008, find the total processing time e (processing ease ()), copy the tentative maximum speed v to oldv, and copy the total processing time e to oldE and oldoldE. The number i of initializing iterations is zero. In addition, ldE was used to determine the improved solution in each of the near-field explorations in step 2016, and ldoldE was used in the loop-out determination in step 2024. Starting from step 20 10, the loop processing for optimizing the station speed and the access sequence in parallel is started. The pull-out of the ring (equal to the end of the process) is when the ring has reached a sufficient number of ring iterations (step 20 1 0), when the total processing time in the ring is completely unchanged (step 2024), and the speed improvement before and after the ring is too When it is small (decided to set 値 T0L) (step 2028). In addition, these escape judgments -56- (53) (53) 200302761 will not matter if you do not use all of them, it is better to use one at least. First, at the start point of the loop (step 2012), a solution S '(processing choose ()) with a possibility of improving the solution is selected from the neighborhood N (S) of S. Proximity use is already shown in Figs. 54 to 59, for example. Next, processing feasibility () is applied to the selected V to calculate the execution probability f of S 1 spoon and the total processing time e (step 20 1 4). In step 2016, it is determined whether S 'is an improved solution. The conditions for improving the solution are feasible (fg0) and the total processing time is shortened (0ldE> e). It is determined that the solution is improved (step 20 1 8), and the solution is updated with V (the tentative optimal access order). For the next determination, the total processing time e is copied into oldE. Then, return to step 2012 to search for an improvement solution again. In the case where it is judged to be a non-improved solution (step 2020), Y is removed from the vicinity, and in step 2022, it is investigated whether there is room for improvement solution exploration in the vicinity. In view of the fact that all the proximate solution processing has ended and there is no room for proximate exploration, it is determined in step 2024 whether the improvement in the loop started from step 2010 has not been performed once. The determination condition is, for example, whether the total processing time oldoldE before the execution of the ring processing is equal to oldE. When it is determined in step 2024 that the improvement is performed once (ie, oldoldE # oldE), the process proceeds to step 2026, and the maximum possible implementation speed v is calculated. Secondly, in step 2028, the maximum speed v of the table before the loop processing is performed can be judged to be hardly improved (for the setting 値 T0L, v-oldV < TOL), and the processing is completed by the loop processing. If the improvement is sufficient, proceed to step 2030. For the next cycle, find the total processing -57- (54) (54) 200302761 time e (processing feasibility ()), copy the tentative maximum speed v to oldv, and process the total processing. Time e is copied to oldE and oldoldE (step 2032), and the loop repeat number I is incremented by one (step 2034), and the process returns to step 2010 to open the loop again. On the other hand, it is determined in step 2024 that the improvement is not performed once (ie, oldoldE = oldE) to complete the process. Furthermore, the XY stage acceleration / deceleration model that smoothly changes the operating speed is optimized by the distribution density of the opening positions determined in step 1 12 of step 2 in FIG. 2 and the opening positions determined in step 1 14 (the speed model determines the process ) Can be performed as follows. Now in a certain processing rectangle, the order of access and the XY stage (constant speed) operating speed of the scanner and the migration on the substrate in the processing area are shown in Figure 63, which is an example of Figure 49. In this example, although it is a processing rectangle in which the scattered state of the opening position is dense and the space is mixed as shown in the figure, the irradiation position of the laser beam of the electric scanner is in the processing area at the spaced position. The end at the earliest start time on the side stops, and it is in a state waiting for the arrival of the XY stage. In the case where the positions of the openings are scattered, the speed of the XY stage is not equal to the speed of the rectangle, but it is preferably a high-speed sparse position. Specifically, as shown in Fig. 64, the operation speed of the XY stage is divided into two. That is to say, the dense program among the consecutive parts (programs) that distinguish the hole position access sequence and the operation speed from the above-obtained hole position access sequence, and the dense program corresponds to the operation speed obtained first, and the other The sparse position corresponds to high-speed operation speed. -58- (55) 200302761 When distinguishing dense programs, for example, select and end of the program as shown in Figure 6-5. That is, first, in step 1000, the candidate at the front selects the initial point of the state of waiting for Ze (the earliest start time-arrival time) is greater; i. In addition, "Look at the program from the beginning" 1 002 The situation where the waiting time is positive does not identify the secret program. Step 1 006 selects the next candidate. Completed at step 1008. Furthermore, it is determined that the final result is reached in step 1004 under the condition that the latest time (the time of completion-the time of illumination) is not smaller than the critical value. On the other hand, for the time when the steps 1 002 and 1004 are negative or not, the time is late-the time when the irradiation is completed), because there is a point more than a certain threshold, so in step 1010,値 Among the small opening positions, let the smallest one be the program. Thus, the dense program can be identified as shown in Fig. 64. Furthermore, the operation speed of the sparse position is maximized, for example, a certain speed that is higher than the setting speed. The operation of driving the position is performed as shown in Figure 66. That is, the dense program distinguished by the method shown in Figure 65 is a program with a thinning, but the acceleration and deceleration time for changing the speed is required. Therefore, each of the thinning programs is directed to step 11 2 Based on the speed of Fang Dahua's operation (the optimization of the order of access to the opening position can be done by using only find_speed () to determine the speed). In addition, the maximum speed of the sparse program is calculated. , (0 moved to 0 in the step, when the end of the step has completed the point of completion, the end of the gestalt (minimum time) is shown as the dense bits at the end and becomes the best for the method, or the acceleration and deceleration -59- (56 ) (56) 2 00302761 Time is also included in the sparse program. If the acceleration and deceleration time is considered to be a constant-speed (high-speed) operation, the speed of the acceleration and deceleration time is lower than the constant-speed operation of the sparse program. The overall program is It can be implemented, so the problem does not occur. As shown in Figure 67 (B), the speed of the above-mentioned thinning program can be increased, and multiple speed correspondence can be performed. That is, the method of step 1 1 4 is applied to the above-mentioned thinning program again. Distinguishing between sparse programs and dense programs can speed up sparse programs. In addition, as a result of the process of optimizing the opening position access sequence and the table operation speed optimization process (step 110), the full point irradiation time is completed later than the latest. It is known that the time is still a certain degree from Fig. 49. There is a margin at the lower limit position of the processing area. In this case, in order to perform processing in the central part of the processing area, it is possible to complete the irradiation at all points earlier than the latest completion time. The processing moments of all the hole positions are internally shifted at the same time. The foregoing step 120 is based on the position of the processing rectangle determined by step 100 and by step 110. Based on the fixed movement speed of the entrance and exit of the processing rectangle, the movement path of the XY stage is optimized as illustrated in Figure 67. The purpose of this optimization is to minimize the movement of a certain (the location of the opening) between the processing rectangles. , That is, the sum of the time when the electrical scanner and the laser beam are at rest. With reference to FIGS. 6 and 8, the operation of the XY stage from a certain processing rectangle to another certain processing rectangle will be described. Regarding the processing rectangles A and B, the operation directions of the XY stage are in two directions. The arrows indicate that the positions on the left and right ends of the machining rectangle are the entrance and exit of the machining rectangle on the migration path toward the machining area. The speed of the XY stage of the entrance and exit Al, A2 of the machining rectangle A is -60 in step no. -(57) (57) 200302761 The optimization decision was made, but the direction of XY stage movement (from right to left or left to right) was not determined. If you consider the symmetry of time and space (perform completely opposite actions), you can reverse the action direction used in step 110, and the speed model and the order of opening position access can also be completely reversed. Therefore, the operation direction of the X-axis platform is also optimized. The operation time of the X-axis stage between the two points is the same as that of the electric scanner, and it is modeled and functional. The case where the arguments of the function are shifted from the state of the coordinates (X !, Υ!), The speed (Vx, Vy) to the state of the coordinates (χ2, γ2), and the speed (Vx, Vy) has 8 arguments (X! , Y !, χ2, γ2, Vx, Vy, Vx, Vy). Let this problem be treated as a trapped salesman problem with restrictions (correctly because it is not necessary to return to the original, so it is a Hamilton path length minimization problem) to capture and solve. Specifically, consider two cities where each processing rectangle has one * end as the entrance and one ^ end as the exit. Moreover, among the lines connecting the cities (called branches), as shown in FIG. 69, the restriction that must be removed for the branches that join the entrance and exit of the adjacent processing rectangle. That is to enumerate such branches alone. The method of constituting the solution may include a heuristic method such as a local search method. Considering both the calculation time and the calculation result, an appropriate method can be used. In addition, in the embodiment described above, not only the processing order of the processing rectangles, but also the processing of the entrance and exit of the processing rectangles must be optimally determined at the same time, so it becomes a complex structure. The entrance and exit (direction of XY stage) is best determined. For example, after optimizing the access sequence of the processing rectangle, and determining the processing of the entrance of the rectangle -61-(58) (58) 200302761, the process can be performed as shown in FIG. 70. That is, firstly, the access sequence of the processing rectangle is optimized in step 1100. In particular, the center of the processing rectangle is used as a city to capture the ‘use the solution of the local exploration by the Hamilton path length minimization problem’ to optimize the access order of the processing rectangle. Secondly, determine the entrance and exit of the processing rectangle. Specifically, in step 1102, for example, the entrance or exit of the processing rectangle to be accessed first is determined such as the left side or the lower side. In addition, let this processing rectangle be a current processing rectangle. Next, in step 1104, among the two entrance and exit candidates of the next processing rectangle, the nearest to the exit of the current processing rectangle is the entrance, and the others are exits. Step 1104 is repeated until the entrance / exit of the processing rectangle of the last access order is determined. Alternatively, step 1104 may be performed as follows. That is, among the two entrance and exit candidates of the next processing rectangle, the nearest to the exit of the current processing rectangle is the temporary entrance (sequential temporary entrance), and the others are temporary exits. About the next two candidates for the next entrance, calculate the cost from the current exit to the next entrance among the two entry-exit candidates for the next processing rectangle, and the cost from the next exit to the temporary entrance in order, so that the lower cost is the next One into □. Or if the number of processing rectangles is K, the number of combinations of the solutions is 2K. If the number of processed rectangles is not large, all the possibilities can be listed completely. Alternatively, even if the number of processed rectangles is large, a local search method or the like may be used to obtain an optimal solution. After deciding the entrance and exit of the processing rectangle, the processing order of -62- (59) (59) 200302761 to optimize the processing of the rectangle is optimized as shown in Fig. 71. That is, firstly, in step 1 200, the entrance and exit of all the processed rectangles are determined. Specifically consider A. If the running direction is left and right, order left, and if it is up and down direction. B. Set randomly for each processing rectangle. Etc. Next, in step 1 202, the access order of the processing rectangle is optimized. Specific processing rectangles are captured one by one as cities. Furthermore, the access sequence of the processing rectangles is optimized by using a locally explored solution by the asymmetric Hamiltonian path length minimization problem. The asymmetry here is because the movement from the processing rectangle A (the exit) to the processing rectangle B (the entrance) and the movement from the processing rectangle B (the exit) to the processing rectangle A (the entrance) have been the entrance and exit of the processing rectangle. Decided, so different. Alternatively, the case where the progress directions of the processed rectangles are the same in the X direction or the Y direction may be performed as shown in FIG. 72 (the case where the progress directions of all the processed rectangles are the X direction). That is, if the initial processing rectangle is the smallest γ-coordinate and there are multiple cases of the same γ-coordinate, the X-coordinate is the smallest. The migration direction on the substrate in the processing area of the first machining rectangle + (positive) direction is set. Moreover, as shown below, they are arranged in order from the smaller Y-coordinate. At this time, the same γ-coordinate makes the movement direction of the processing area the same from left to right. That is, from the leftmost to the right, or vice versa. On the other hand, when the Y coordinate is different, the current processing rectangle is moved to the right in the processing area on the substrate. If the left end of the next processing rectangle is more left than the right end of the current processing rectangle, the migration direction is not changed. The migration direction is reversed. Or use -63- (60) (60) 200302761 or the opposite. In this way, the processing rectangle moves from a small Y-coordinate to a large one, so that the moving path of the processing area on the substrate is meandered and connected without waste. In addition, in the aforementioned embodiment, as shown in FIG. 2, the processing rectangle of the opening position is divided first, and then the access order optimization of the opening positions in each processing rectangle and the XY stage operation speed are optimized, and finally The XY stage operation path is optimized. However, in the second embodiment shown in FIG. 73, firstly, the order of opening positions of the entire substrate is optimized in step 1300, and secondly, in step 1301, the XY stage control can be effectively set. The target position (specified position during console operation) determines the XY stage operation model. When the order of accessing the opening positions of the entire substrate in step 1 300 is optimized, for example, the problem is captured as a symmetrical traveling salesman problem in which the cost between two points AB is moved by A and the situation where B is moved. For example, optimization can be performed by using a known technique such as an algorithm of L i η-K ernigha η. On the other hand, for the XY stage motion model decision at step 1310, the speed or acceleration plan of the XY stage is very difficult because the stage is not limited to linear motion. Therefore, the target position is determined for the X-axis stage for scanning by the electric scanner and processing by laser irradiation between the steps toward the target. In this method, in order to ensure the action that can be performed, the target position of the X-axis stage is set as shown in Fig. 74. The first consideration is to distinguish the hole positions with rectangular groups of the same size as the processing area. Regarding the opening position determined by the aforementioned step 1 300 (the order of P :, P2, .. is added to the access sequence), in step 1312, the head P! Of the access sequence is -64- (61) (61) 200302761. Point (i = l). Moreover, the group is represented by {D !, D2 ... 丨, and the subscript k is set to 1 〇 Moreover, as shown in steps 13 14 to 1 320, by repeating the next process until the last hole position is reached, It is divided into opening positions by a rectangular group of the same size as the processing area. That is, in step 1314, regarding j (S 1), find the largest j that all images are contained in a rectangle of the same size as the processing area, and in step 1 3 1 6 let {Pi ... Pj} be a group Dk. Next, in step 13 18, the next attention point i is made j + 1, and k is increased. Next, in steps 1 to 22, the target position of the XY stage is set as follows. That is, for the point set {Pi ... Pd divided into Dk by the group, let the enlarged center position of {Pi ._. Pd be the target position. Here, the center of the enlargement is that the {minimum 値, maximum 値} in the X direction and Y direction in the instruction {Pi… Pj} are UnHn,
Xmax}、{ymin、ymax} ’ 以座標{ (Xmin + Xmax) /2、 C Y ni i η + y…）/2}表示的位置。 使用如此設定的目標位置，在步驟1 330〜1350實現可 實行的台非停止加工。 即在步驟1 3 30關於被群區分成Dk的點集合{Pi... P]， 在XY台朝Dk的目標位置動作的過程中，依次進行藉由 電掃描器的定位與藉由雷射照射的加工。 在步驟1 340隨著所有的點{Pi.·· P]的加工完成，在步 驟1 3 50使XY台移動到下一目標位置。 此外，在步驟1 322因令擴大的中心位置爲XY台的 -65- (62) (62)200302761 目標位置，故簡單，但未必是最佳的設定方法。 因此，說明其他的設定方法。首先，參照圖75針對 成爲目標位置的候補的區域來說明。在點集合{Pi··. Pj}之 中，令X方向、Y方向的（最小値、最大値}分別爲、Xmax}, {ymin, ymax} ’are represented by coordinates {(Xmin + Xmax) / 2, C Y ni i η + y ...) / 2}. With the target position set in this way, the non-stop machining that can be performed in steps 1 330 to 1350 is realized. That is, in step 1 3 30, regarding the point set {Pi ... P] divided into groups by Dk, in the process of the XY stage moving toward the target position of Dk, the positioning by the electric scanner and the laser are sequentially performed. Irradiated processing. At step 1 340, as all the points {Pi .. · P] are processed, at step 1 3 50, the XY stage is moved to the next target position. In addition, since the expanded center position in step 1 322 is the -65- (62) (62) 200302761 target position of the XY stage, it is simple, but it may not be the best setting method. Therefore, another setting method will be described. First, an area to be a candidate for the target position will be described with reference to FIG. 75. In the point set {Pi ··. Pj}, let (min 値, maximum 値) in the X direction and Y direction be,
Xmax}、{ymin、ymax}。點的 大爲{Xmin、ymin}、{X_max、 ymin}、U…、yn-d、的四點包圍的矩形，惟若 令這四點均收容的加工區域的大小的矩形中心爲目標位置 ，則可照射。因此，在加工區域爲正方形時的例子若令加 工區域的一邊爲d則成爲Dk的目標位置的候補的矩形Ck 的頂點座標爲 {Xmax-d/2 Λ ymax-d/2} (Xmin + d/ 2、ymax-d/2}、 {Xmin + d/ 2、ymin + d/ 2}、 (Xmax-d/2、ymin + d/ 2} 0 XY台的動作因比電掃描器的動作慢，故少者較佳。 因此’取代令擴大的中心位置爲目標位置，如圖75所示 地設定也可以。此處，對於表示Dk的目標位置使用Vk。 連結C!的四頂點與C2的四頂點的組合之中，設定最 近的組合的頂點爲V:、V2。 關於i>2，令C!的四頂點之中最接近Vu的點爲Vi。 或者如圖77以及圖78 (槪念圖）所示，設定成像由 Vm到Vi的距離與由1到Vi + 1的距離的和爲最小的^也 可以。即首先在連結Ci的四頂點與C2的四頂點的組合， 設定最近的組合的頂點之中Ci的頂點爲Vi。 -66- (63) (63)200302761 其次，關於i - 2暫定地 甲、暫設定Vi爲C!的四頂點之內，距Vi^最近的頂 點。 乙、暫設定Vi + 1爲Cu!的四頂點之內，距Vm最近的 頂點。 如圖77所示，爲了說明起見取Vi爲原點，令在 第四象限具有（若考慮對稱性的話則一般性不會喪失）。 如圖77 ( a )所示，依照Vi+ 1位於哪一象限進行如以 下的處理。 即 例子A:位於第一象限時，設定[vi + 1]爲與V;同一， 例子B1··位於第二象限時，令Vi + 1關於Y軸成對稱 的位置爲[V i + 1 ]， 例子B2:位於第四象限時，令Vi + 1關於X軸成對稱 的位置爲[V i + 1 ]， 例子C:位於第四象限時，在暫時決定Vi的位置確定 。以下的處理不進行。 其次如圖77 ( b)所示，令連結位於第一象限的[Vi + 1] 與位於第四象限的[Vw]的直線爲L。根據矩形Ci與直線L 的交叉關係如以下決定Vi。 即直線L爲矩形Ci與 甲、無交叉且L的Y切片爲負的情形，確定Ci右下 的頂點P爲Vi。Xmax}, {ymin, ymax}. The point is a rectangle surrounded by four points of {Xmin, ymin}, {X_max, ymin}, U ..., yn-d, but if the center of the rectangle of the size of the processing area accommodated by these four points is the target position, It can be irradiated. Therefore, in the case where the processing area is a square, if the side of the processing area is d, the vertex coordinate of the rectangle Ck which is a candidate for the target position of Dk is {Xmax-d / 2 Λ ymax-d / 2} (Xmin + d / 2, ymax-d / 2}, {Xmin + d / 2, ymin + d / 2}, (Xmax-d / 2, ymin + d / 2) 0 The operation of the XY stage is slower than that of the electric scanner Therefore, the less is better. Therefore, 'replace the enlarged center position as the target position, as shown in FIG. 75. Here, Vk is used for the target position representing Dk. The four vertices of C! And C2's Among the combinations of four vertices, the nearest combination is set to V: and V2. For i > 2, let the point closest to Vu among the four vertices of C! Be Vi. Or as shown in Figure 77 and Figure 78 (槪 念As shown in the figure), it is also possible to set the sum of the distance from Vm to Vi and the distance from 1 to Vi + 1 to be the smallest ^. That is, first set the combination of the four vertices of Ci and the four vertices of C2 to set the nearest Among the combined vertices, the apex of Ci is Vi. -66- (63) (63) 200302761 Second, about i-2 tentative lands, temporarily set Vi as the four vertices of C !, the top point closest to Vi ^ B. For the time being, set Vi + 1 to be the vertex closest to Vm within the four vertices of Cu !. As shown in Figure 77, take Vi as the origin for the sake of illustration, so that in the fourth quadrant (if symmetry is considered (The generality will not be lost). As shown in Figure 77 (a), the following processing is performed according to which quadrant Vi + 1 is located. That is, Example A: When it is in the first quadrant, set [vi + 1] to be equal to V ; Same, Example B1 ·· When the second quadrant, make Vi + 1 symmetrical about the Y axis is [V i + 1], Example B2: When in the fourth quadrant, make Vi + 1 symmetrical about the X axis The position of is [V i + 1]. Example C: When it is located in the fourth quadrant, the position of Vi is temporarily determined. The following processing is not performed. Next, as shown in Figure 77 (b), the link is located in the first quadrant. The straight line between [Vi + 1] and [Vw] in the fourth quadrant is L. According to the intersection relationship between the rectangle Ci and the line L, Vi is determined as follows. In the negative case, it is determined that the vertex P at the lower right of Ci is Vi.
乙、有交叉且L的Y切片爲負的情形，確定Ο與L -67- (64) (64)200302761 的交點Q 1、Q2或位於其兩點間的任意位置爲Vi。 丙、有交叉且L的Y切片爲正的情形，確定◦與L 的交點Rl、R2或位於其兩點間的任意位置爲。 丁、無交叉且L的Y切片爲正的情形，確定Ci左上 的頂點S爲Vi。 如此若確定的話，由Vu!到Vi的距離與由Vi到Vi + 1 的距離的和爲最小。 上述兩個設定方法並非都是探索完全最佳的目標位置 的方法，惟可以少的計算時間得到良好的計算結果。或者 ，定義由表示各目標位置的矩形選擇幾個位置的目標位置 的候補，由這些候補的組合求佳的組合的組合最佳化問題 ，例如也能使用局部探索法等的探試求佳的解。 此外，對XY台決定目標位置，在朝目標位置動作之 間採用藉由電掃描器的掃描以及藉由雷射照射的加工的方 法的情形，實際上依照XY台的機械特性有精度或追蹤性 等發生問題。因此可考慮如第三實施形態，XY台的一方 在加工中停止，僅在X方向或Y方向的一方藉由使用決 定XY台的目標位置的方式的台非停止加工來實現。即倂 用已經敘述的加工矩形分割，如圖79所示分割製程。 即在步驟1400決定開孔位置的加工矩形分割，在步 驟1 4 1 0最佳化加工矩形內的開孔位置訪問順序’在步驟 1 4 20進行XY台的動作模型即進行加工矩形內的開孔位置 的群化，設定各群的XY台控制目標位置。 前述步驟1400爲與圖2的步驟100同一。 -68- (65) 200302761 步驟1410可使用在圖44說明的連結初期路徑設定 以及始點與終點的枝必須切斷的附有限制的巡迴售貨員 題的解法。 而且，關於步驟1420可使用與圖73的步驟1310 樣的XY台動作模型最佳化。 如此，進行朝X或Y的一方向的XY台動作，藉由 該一列的加工完了的話加工下一列的順序佳的加工，可 現定位精度佳的台非停止加工。 而且，在朝X或γ的一方向的xy台動作中，藉由 在開孔位置存在的位置進行加工矩形分割，可實現有效 台非停止加工。 而且，在朝X以及Y的一方向的XY台動作中，藉 試行X以及Y的加工矩形分割的兩方，選擇佳者，可 現有效的台非停止加工。 而且，在朝X以及Y的一方向的XY台動作中，藉 試行X以及Y的加工矩形分割的兩方，選擇產生的加 矩形分割之內XY台的動作距離小者，進行XY台的動 距離最小的加工矩形分割，可實現有效的台非停止加工 而且，因藉由X或Y軸方向的最佳加工矩形分割 可僅限定XY台的動作位置於具有開孔位置的位置，故 實現加工高速化。 而且，藉由考慮開孔位置的疏密的加工矩形分割， 的位置在疏的位置，密的位置在密的位置特定加工矩形 藉由在疏的位置以快的速度，在密的位置以慢的速度進 法 問 像 實 僅 的 由 實 由 工 作 0 5 可 疏 5 行 -69- (66) (66)200302761 加工，可實現加工高速化。 而且，在藉由XY台的驅動與電掃描器的掃描都進行 的定位中，配合開孔位置於XY台動作方向以適當的寬度 群區分，在各群中分類於與XY台進行方向垂直的方向， 藉由連接成蛇行狀可實現在浪費少的路徑的加工。 而且，在藉由XY台的驅動與電掃描器的掃描都進行 的定位中，藉由依照該加工矩形的開孔位置密度設定上述 適當的寬度，可實現在更無浪費的路徑的加工。 而且，在藉由XY台的驅動與電掃描器的掃描都進行 的定位中，藉由預先模型化電掃描器的動作時間，使在模 型上的電掃描器的動作時間的總和爲最小而重複改善開孔 位置的訪問順序，可實現動作時間更短的加工。 而且，在藉由對XY台的某方向的驅動與電掃描器的 掃描都進行的定位中，依次設定XY台的控制目標位置， 對於像在正在朝目標位置驅動中進行照射的形態的情形， 藉由事前計劃目標位置，可有效地進行台非停止加工。 在前述形態中，藉由依次進行令將開孔位置由最初訪 問者到依次收容於與掃描區域相同大小的矩形爲止爲一個 群的處理，藉由決定目標位置的方法可設定佳的目標位置 〇 關於前述目標位置，藉由令表示屬於開孔位置的群的 所有點收容於掃描區域的矩形的中心位置的矩形區域爲目 標位置的候補，可增加用以減少XY台驅動量的融通性。 在最佳化遍及前述目標位置的路徑的過程中，藉由令 -70- (67) (67)200302761 下一目標位置爲距前一目標位置最近的位置，可決定佳的 目標位置。 或者’在最佳化遍及前述目標位置的路徑的過程中， 藉由在決定下一目標位置時，暫時決定下下一個目標位置 爲距則一目標位置最近的位置，令最小化從前一目標位置 的成本與到下下一個目標位置的移動成本的和的位置爲目 標位置，可決定佳的目標位置。 或者’在最佳化遍及前述目標位置的路徑的過程中， 決定幾個各目標位置的候補，定義求像由目標位置的候補 的組合最小化移動成本的總和的組合的組合最佳化問題， 藉由局部探索法等的手法可決定優良的目標位置。 而且，在藉由XY台的驅動與電掃描器的掃描都進行 的定位中，對於預先設定XY台的速度、加速模型等，依 設定而驅動，在驅動中進行照射的形態的情形，藉由事前 較佳地計劃XY台的速度、加速模型等，可有效地進行台 非停止加工。 或者，在藉由XY台的驅動與電掃描器的掃描都進行 的定位中，對於預先設定XY台的速度、加速模型等，依 設定而驅動，在驅動中進行照射的形態的情形，藉由事前 較佳地計劃XY台的等速的動作速度，可削減加工計劃時 間，且可有效精度佳地進行台非停止加工。 在前述形態中，藉由對分割開孔位置的各加工矩形， 決定動作速度，可提高加工速度。 在前述形態中，在像開孔加工中以等速使XY台動作 -71 - (68) 200302761 的實行形式中，藉由當進行計劃時在模型上決定等速動作 中的最早照射開始時刻以及最遲照射完了時刻，求在該時 序（Timing)內如收容照射開始時刻以及照射完了時刻的 最大速度的方法，可求等速動作的最大界限。 前述最大速度藉由首先決定可實行速度與不可實行速 度’關於間隔的速度調查實行可能性，若可實行則逐次更 新可實行速度，若不可實行則逐次更新不可實行速度，可 有效地求實行可能速度的最大界限。 在前述形態中，在像開孔加工中也能使XY台可變動 作的實行形式中，藉由首先求等速動作的最大値，特定限 制該動作的密的程序，關於疏的程序更高速化，可比等速 動作還使XY台的動作更高速化。 在前述形態中，在像開孔加工中也能使XY台可變動 作的實行形式中，藉由由進行等速動作的程序之中切割密 的程序與疏的程序，對疏的程序多重化使動作速度高速化 的處理，可使XY台的動作更高速化。 在前述同時最佳化中，藉由對某開孔位置訪問順序與 某XY台速度作成調查實行可能性的函數，若對某速度可 實行的話則提高實行可能速度的上限，若爲不可實行的話 則改善開孔位置訪問順序，使實行可能而探索可實行的點 訪問順序與χγ台速度，使若可實行的話則提高實行可能 速度的上限，若爲不可實行的話則降低實行不可能速度的 下限，可有效地求出等速動作的最高速度。 或者，在前述同時最佳化中，使用對某開孔位置訪問 -72- (69) (69)200302761 順序與某XY台速度調查實行可能性的函數，作成最大化 對某開孔位置訪問順序的ΧΥ台等速動作速度的函數，若 其他訪問順序中的動作速度比某開孔位置訪問順序中的動 作速度還快的話，藉由儘可能重複更新可實行的動作速度 以及點訪問順序的處理，可有效地求出等速動作的最高速 度。 而且，關於一邊使ΧΥ台動作一邊進行照射加工時的 開孔位置存在的加工矩形，藉由最佳地決定使訪問複數個 加工矩形的順序以及使加工矩形上動作的方向（動作路徑 )，可高速化加工速度。 前述決定法藉由若加工矩形的動作方向被左右或上下 均勻化的話，使ΧΥ台的動作路徑由基板的端成爲蛇行狀 而設定，可高速化加工速度。 或者，前述決定法藉由首先令各加工矩形的中心位置 爲各都市的座標，適用巡迴售貨員問題或哈密頓路徑長最 小化問題，以決定加工矩形訪問順序，其次在決定各加工 矩形上的動作方向時，以藉由重複選擇入出口的候補兩個 以內由前一加工矩形的出口的移動成本小者來決定注目的 加工矩形的下一加工矩形的入口之方法，可決定佳的χγ 台的動作路徑。 而且，藉由首先令注目的加工矩形的下下一個加工矩 形的入出口的候補兩個以內由注目的加工矩形的出口的移 動成本小者爲暫時的入口’重複選擇注目矩形的下一個加 工矩形的入出口的候補兩個以內由注目矩形的入口下一個 -73- (70) (70)200302761 入口候補的移動成本與下一個出口候補與暫時的入口的移 動成本的和小者，可決定佳的X Y台的動作路徑。 而且，各加工矩形的入口候補有兩個，加工矩形的數 目少的情形藉由採取像關於所有的可能性求移動成本的總 和，選擇最小者的方法，可決定最佳的入出口的選擇。 而且，各加工矩形的入口候補有兩個，藉由局部探索 法等的手法，可決定優良的入出口的選擇。 前述決定法首先決定各加工矩形的XY台動作方向， 以這些加工矩形的入出口爲基礎，根據非對稱售貨員問題 或哈密頓路徑長最小化問題的解法，可決定佳的XY台的 動作路徑。 前述決定法令各加工矩形的入出口爲兩個都市來捕捉 ，藉由使用在連結入出口的枝必須去除的限制的基礎下的 對稱巡迴售貨員問題或哈密頓路徑長最小化問題的解法來 最佳化，可同時決定訪問加工矩形的順序以及加工區域爲 移動於加工矩形上的方向。 而且，在進行台非停止加工時，藉由首先定義開孔位 置訪問路徑爲令基板全體的各開孔位置爲一個都市的對稱 巡迴售貨員問題或哈密頓路徑長最小化問題，然後決定可 實行的XY台的目標位置，可實現台非停止加工。 而且，藉由分割基板全體的開孔位置爲一邊（進行邊 )爲加工區域的寬度以下，他邊（停止邊）爲加工區域的 寬度以上的平行於XY軸的幾個加工矩形，開孔加工中藉 由使XY台進行朝X方向或Y方向的同一方向的單純的動 -74- (71) (71)200302761 作，若加工完成的話高速地移動到下一個加工矩形，可有 效地進行速度、加速度模型的計劃。 特別是對於如圖6或圖1 0決定上述加工矩形分割中 的各加工矩形，使XY台動作路徑成爲蛇行狀而設定的情 形，可得到非加工中的情形的動作少的佳的路徑。 而且，在以上述加工矩形分割得到的各加工矩形內， 無開孔位置的位置如圖8所示沿著進行邊存在長距離的情 形，藉由在該位置的前後或左右二等分掃描矩形，可識別 開孔位置不存在的位置爲遍及加工矩形間的部分，可使 XY台動作速度計劃容易。 而且，藉由前述加工矩形分割如圖10全加工矩形的 進行邊與停止邊的方向統一，加工矩形的位置如圖丨i ( B )，進行像互相段一致的配置的加工矩形分割，使XY台 動作路徑最佳化容易，可精度佳地移動XY台，可進行實 用的XY台動作速度計劃。 即前述段一致的配置使用最短路徑問題的解法最小化 段的個數，可進行有效的加工矩形配置。 而且，前述段一致的配置藉由使加工矩形內的開孔位 置接近加工區域的中心而微調整各加工矩形的位置，可實 現精度佳的加工。 再者，前述段一致的配置藉由選擇最佳化的結果段的 數目爲最小的解之中，使沿著各加工矩形內的段的某方向 的開孔位置不存在的間距爲最小者，可令各加工矩形內的 開孔位置配置對加工高速化爲有效者。 -75- (72) (72)200302761 而且，藉由如圖6或圖1 〇以加工矩形的進行邊與停 止邊的方向統一者計劃前述加工矩形分割時，針對X方向 、Y方向的兩方試行，選擇佳者，可縮短加工時間。 前述佳者的選擇基準是藉由選擇例如加工矩形的進行 邊的長度的總和小者，可得到無浪費的X γ台路徑。 或者，前述佳者的選擇基準藉由調查各加工矩形的疏 密的均勻度，可使XY台的動作速度對各加工矩形大致一 定，可實現無浪費的加工。 再者，爲了調查前述各加工矩形的疏密的均勻度，藉 由沿著加工矩形的進行邊作成累積頻率分布，可容易且正 確地檢測如圖1 5 ( B)可見到的疏密的不均。 而且，藉由如圖6或圖10以加工矩形的進行邊與停 止邊的方向統一者計劃前述加工矩形分割時，如圖21所 示針對X方向、Y方向的兩方試行，一旦決定加工矩形位 置後，藉由這些加工矩形的佳的組合決定加工矩形的配置 ，可縮短加工時間。 前述佳的組合的基準是藉由選擇例如加工矩形的進行 邊的長度的總和小的組合，可得到無浪費的χγ台路徑。 或者，前述佳的組合的基準是藉由選擇各加工矩形的 疏密爲均勻的組合，使XY台動作速度對各加工矩形大致 一定，可實現無浪費的加工。 而且，前述加工矩形分割判定是否有開孔位置的疏密 ，疏的位置與密的位置存在的情形依照疏密的程度群區分 開孔位置，關於最疏群以外的群的開孔位置’針對有規則 -76- (73) (73)200302761 性的位置藉由先確定加工矩形位置，如圖27所示可使加 工時的開孔位置的密度一樣化，可提高加工效率。 此外，爲了判定前述開孔位置的疏密，藉由分割基板 全體成格子狀，求各胞（cell )的開孔位置的個數，作成 度數圖，可容易且正確地檢測開孔位置的疏密。 而且，在檢測前述開孔位置的各群的規則性時，分割 基板全體成格子狀，調查各胞的開孔位置的疏密，依照該 資訊如圖26所示，藉由以被開孔位置疏的胞的環包圍的 覆蓋包含開孔位置密的胞的胞群的任一個的胞集合矩形來 被覆全開孔位置，可容易且正確地檢測規則性。 而且，前述加工矩形分割判定根據開孔位置的分布狀 態或資料檔案的書式，是否可二等分開孔位置爲大多數群 與少數群，如圖31所示對於可二等分的情形，關於大多 數群的開孔位置，藉由對具有規則性的位置確定加工矩形 位置，可容易實現在段一致的加工矩形配置方法因少數開 孔位置的存在而困難的加工矩形配置，可提高加工效率。 此外，爲了檢測前述開孔位置的大多數/少數群的規 則性，分割基板全體成格子狀，調查各胞的開孔位置的有 無，依照該資訊如圖26所示，藉由以被無開孔位置的胞 的環包圍的覆蓋包含有開孔位置的胞的胞群的任一個的胞 集合矩形來被覆全開孔位置，可容易且正確地檢測規則性 〇 再者，關於所得到的胞集合矩形的每一個，藉由調查 位於矩形外周側的胞內的最外側的點，可更正確地特定前 -77- (74) (74)200302761 述開孔位置的擴大，可容易地檢測規則的配置。 而且，前述最疏群以外的群以及大多數群的規則性藉 由令成直線狀配置或格子狀配置，可實現可使XY台在同 一方向動作的掃描矩形分割。 而且，由前述最疏群以外的群的複數個胞集合矩形以 及大多數群的複數個胞集合矩形檢測直線狀配置，以覆蓋 胞集合矩形來配置表示加工位置的加工矩形時，如圖37 (A)所示，以令直線的延伸方向爲進行邊的一個或複數 個分割矩形被覆直線狀配置，若進行直線狀配置的胞集合 矩形的間隔大到某種程度的話，如圖37 ( B )所示’藉由 在該位置的前後二等分，可識別開孔位置不存在的位置爲 遍及加工矩形間的部分，可使XY台動作速度計劃容易。 而且，進行直線狀配置的胞集合矩形的與直線的延伸 方向垂直的方向的大小如圖3 7 ( C )所示’若比加工區域 的大小還大的話，藉由配置直線狀配置成複數個加工矩形 ，可容易得到加工矩形。 再者，藉由如前述的複數個加工矩形進行被覆時’在 加工矩形的內部如圖1 5 ( A)所示’在開孔位置被二等分 的位置產生的情形如圖3 7 ( D )所示’藉由微調整加工矩 形的中心位置，可令各加工矩形內的開孔位置配置對加工 高速化爲有效者。 而且，由前述最疏群以外的群的複數個胞集合矩形以 及大多數群的複數個胞集合矩形檢測格子狀配置’以覆蓋 胞集合矩形來配置表示加工位置的加工矩形時’如圖3 8 -78- (75) (75)200302761 所示，以一個或複數個平行的加工矩形被覆格子狀配置， 關於格子狀配置的X方向、Y方向的每一個，若胞集合矩 形的間隔大到某種程度的話，如圖3 8所示分斷矩形群。 此時，以覆蓋格子狀配置的胞集合矩形的方式在配置一個 或複數個平行的加工矩形時的進行邊的方向的設定方法爲 甲、藉由XY台的安裝（上段、下段），藉由在成爲 上段台的方向設定進行邊，可得到XY台動作追蹤性、精 度等佳的加工中的χγ台動作路徑。 乙、藉由以加工矩形片數少即可來設定，可降低遍及 加工矩形間的次數。 丙、藉由以進行邊的長度的總和小者或胞集合矩形彼 此之間的距離的總和小者來設定，可降低在加工中開孔位 置不出現的時間。 丁、藉由模型化電掃描器或XY台的動作，實際計算 總加工時間，選擇佳者，可得到有效的加工矩形。 而且，前述加工矩形分割首先判定開孔位置的疏密， 對於存在疏的位置、密的位置的情形，依照疏密的程度群 區分開孔位置，關於最疏群以外的群的開孔位置對有規則 性的位置確定加工矩形位置，無的位置剩餘，其次判定根 據開孔位置的分布狀態或資料檔案的書式，是否可二等分 開孔位置成大多數群與少數群，對於可二等分的情形，關 於大多數群的開孔位置，藉由對具有規則性的位置確定加 工矩形位置，無的位置剩餘，最後確定剩餘的位置的加工 矩形位置，使開孔位置的疏密一致，可得到可無浪費地加 -79- (76) (76)200302761 工圖案形狀的理想的加工矩形分割，可使加工效率化。 而且，在前述剩餘位置的加工矩形位置確時序（ Timing)，藉由採用如圖6、圖10、圖17、圖21所不的段 一致的加工矩形配置，可使加工效率化。 而且，在前述剩餘位置的加工矩形位置確時序（ Timing)，剩餘的開孔位置的存在位置少的情形藉由適用 台停止加工，可使加工效率化。 此外，進行前述台非停止加工時的台停止加工的倂用 可藉由以台停止加工爲台非停止加工的一部分來捕捉，可 以同一控制系來實現，使設計容易。 此外，在前述實施形態中雖然令高速定位手段爲電掃 描器，令低速定位手段爲X Y台，惟定位手段的種類或組 合並非限定於此，例如像申請人於日本特開2 0 0 0 - 7 1 0 8 9 或日本特開2 0 0 0 - 3 3 4 6 3 7號提出的組合線性馬達X γ台與 高速加工頭的混合加工系統（所謂的篩切割系統（screen cut system )或快速切割系統（flash cut system ))也可以 ο 而且，顯然地適用對象也不限定於進行點狀加工的雷 射開孔機’進行線狀加工的雷射切斷機、使用日本特開平 11-149317藏所S載的兩頭雷射加工機或雷射束以外的加 工手段的一般的加工機（例如利用機械式鑽頭的開孔裝置 )’藉由同時驅動兩個以上的定位裝置，可高速化定位者 全般都同樣地可適用。 -80 - (77) (77)200302761 【產業上的可利用性】 如果依照本發明，具有在加工的事前階段中有效地進 行加工用的計劃爲可能，可實現定位精度佳的低速定位手 段非停止加工，可縮短加工時間的優良的功效。 【圖式簡單說明】 圖1是顯示本發明的適用對象的一例的雷射開孔機的 主要部位構成的前視圖。 圖2是顯示藉由本發明的加工計劃的第1實施形態的 處理順序的流程圖。 圖3是顯示藉由本發明的加工矩形分割最佳化的槪念 的俯視圖。 圖4是同樣地顯示加工矩形內的開孔位置訪問順序以 及XY台動作速度最佳化的槪念的俯視圖。 圖5是同樣地加工矩形動作路徑最佳化的槪念圖。 圖6是同樣地加工矩形分割的一例的槪念圖。 圖7是同樣地顯示最佳分割困難的基板的例子的俯視 圖。 圖8是同樣地分割一個加工矩形成複數個的手法的槪 念圖。 圖9是同樣地說明分割加工矩形時所需的最小距離的 槪念圖。 圖1 〇是同樣地加工矩形分割的其他例的槪念圖。 匱I 11是同樣地顯示段一致的配置的例子的俯視圖。 -81 - (78) (78)200302761 圖1 2是同樣地顯示段一致的配置的加工矩形分割的 順序的流程圖。 圖1 3是同樣地顯示段一致的配置的加工矩形分割的 定式化的槪念圖。 圖14是同樣地顯示段一致的配置的加工矩形分割的 定式化的槪念的線圖。 圖1 5是同樣地顯示並非有效的加工矩形分割的例子 的俯視圖。 圖1 6是同樣地顯示段一致的矩形配置的加工矩形分 割的情形的加工計劃的順序的流程圖。 圖1 7是同樣地顯示藉由縱方向優先與橫方向優先的 分割之後，選擇佳者的手法的槪念的俯視圖。 圖1 8是同樣地顯示加工矩形內開孔位置的疏密與XY 台動作速度的關係的例子的俯視圖。 圖1 9是同樣地顯示藉由加工矩形的疏密的均勻度的 分割方向選擇的順序的流程圖。 圖20是同樣地調查加工矩形的疏密的一樣性用的累 積頻率圖的說明圖。 圖21是同樣地顯示藉由縱方向優先與橫方向優先的 組合的加工矩形分割的例子的俯視圖。 圖22是同樣地顯示由複數個圖案形狀（大多數）與 其他少數的開孔位置構成的開孔位置的配置例的俯視圖。 圖23是顯示開孔位置的加工矩形分割的製程的流程 圖0 -82- (79) (79)200302761 圖24是顯示前述加工矩形分割製程中的密的位置的 加工矩形配置處理的流程圖。 圖25是同樣地顯示大多數位置的加工矩形配置處理 的流程圖。 圖26是藉由開孔位置的有無的點存在位置的加工矩 形化的說明圖。 圖27是顯示圖24的處理中的格子分割以及格子的識 別處理的流程圖。 圖28是開孔位置的疏/密的階層區分的槪念圖。 圖29是開孔位置的密的格子與疏的格子的分布的例 子的俯視圖。 圖30是顯示圖25的處理中的格子分割以及格子的識 別處理的流程圖。 圖31是藉由本發明的開孔位置的大多數/少數的群區 分的槪念圖。 圖32是顯示圖24、圖25的處理中的格子群矩形化 處理的流程圖。 圖33是顯示圖24的處理中的格子群矩形位置的微調 整處理的流程圖。 圖34是顯示圖25的處理中的格子群矩形位置的微調 整處理的流程圖。 圖35是顯示藉由圖24、圖25的處理中的格子群矩 形的加工矩形的被覆（加工矩形的配置）處理的流程圖。 圖3 6是藉由本發明的規則的排列的配置的識別法的 -83- (80) (80)200302761 說明圖。 ® 37是同樣地關於直線狀配置的被覆的說明圖。 3 8是同樣地關於格子狀配置的被覆的說明圖。 ® 39是同樣地顯示開孔位置訪問順序決定與XY台 # ^動作最大化的順序的例子的流程圖。 ® 40是同樣地顯示其他例子的流程圖。 ® 4 1是同樣地顯示再其他例子的流程圖。 ® 42是同樣地顯示再其他例子的流程圖。 ® 43是同樣地顯示開孔位置訪問順序決定與XY台 等Μ動作最大化的順序的例子的流程圖。 圖44是顯示圖43的具體例的流程圖。 圖45是顯示開孔位置訪問的初期路徑的說明圖。 圖46是同樣地顯示分割寬與平均移動量的關係例子 的線圖。 圖47是同樣地顯示對某開孔位置訪問順序算出可實 行的最大速度的順序的流程圖。 圖48是同樣地以二分法求出ΧΥ台等速最大速度的 手法的槪念圖。 圖49是同樣地實行可能性的說明圖。 圖50是同樣地求出最早開始以及最遲完了的時刻的 計算例的說明圖。 圖5 1是同樣地顯示兩點間的移動所花的時間的模型 例的說明圖。 圖5 2是同樣地求出低速側ΧΥ台速度的順序的流程 -84 - (81) (81)200302761 圖。 圖53是同樣地求出高速側XY台速度的順序的流程 圖。 圖54是同樣地以比例法求出ΧΥ台等速最大速度的 手法的槪念圖。 圖55是同樣地藉由2交換近傍的路徑更新的說明圖 〇 圖56是同樣地藉由前方Or-opt近傍的路徑更新的槪 念圖。 圖57是同樣地藉由後方Or-opt近傍的路徑更新的槪 念圖。 圖5 8是同樣地藉由插入近傍（無反轉）的路徑更新 的槪念圖。 圖59是同樣地藉由插入近傍（有反轉）的路徑更新 的槪念圖。 圖60是同樣地藉由交叉交換近傍的兩條路徑更新的 槪念圖。 圖61是同樣地藉由2-opt*近傍的兩條路徑更新的槪 念圖。 圖62是同樣地顯示加工位置訪問順序最佳化以及低 速定位手段等速動作最大化製程的具體例的流程圖。 圖63是同樣地顯示實基板的某加工矩形的某開孔位 置訪問順序的X Y台與電掃描器的動作曲線的例子的線圖 -85- (82) (82)200302761 圖64是同樣地速度可變對應的槪念圖。 圖65是同樣地顯示區分密的程序的順序的流程圖。 圖66是同樣地顯示速度可變對應的速度的線圖。 圖67是同樣地速度多重化的槪念圖。 圖68是同樣地加工矩形間的動作的說明圖。 圖69是同樣地加工矩形動作路徑最佳化的說明圖。 圖70是同樣地顯示加工矩形的訪問順序最佳化與入 出口最佳決定的順序的流程圖。 圖71是同樣地顯示加工矩形的入出口決定與訪問順 序最佳化的順序的流程圖。 圖72是同樣地加工矩形動作路徑的簡單的設定法的 槪念圖。 圖73是同樣地顯示加工計劃的第2實施形態的處理 順序的流程圖。 圖74是同樣地其他構成例的槪念圖。 圖75是同樣地成爲目標位置的候補的區域的說明圖 〇 圖76是同樣地XY台的目標位置的設定的槪念圖。 圖77是同樣地說明圖。 圖78是同樣地槪念圖。 圖79是顯示藉由本發明的加工計劃的第3實施形態 的處理順序的流程圖。 【主要元件對照表】 -86 - (83) (83)200302761 1 0 :工件 12: XY 台 20:雷射光線 22:第一電掃描器 23、25:旋轉鏡 24:第二電掃描器 26: f- 0透鏡B. In the case where there is a cross and the Y slice of L is negative, it is determined that the intersection Q1, Q2 of 0 and L -67- (64) (64) 200302761, or any position between the two points is Vi. C. In the case where there is a cross and the Y slice of L is positive, it is determined that the intersection points R1 and R2 with L or any position between the two points is. In the case of D, no intersection, and the Y slice of L is positive, it is determined that the vertex S on the upper left of Ci is Vi. If so, the sum of the distance from Vu! To Vi and the distance from Vi to Vi + 1 is the smallest. The above two setting methods are not both methods to explore the completely optimal target position, but can obtain good calculation results with less calculation time. Alternatively, a combination optimization problem that selects several target positions from a rectangle representing each target position and selects a combination of these candidates to determine the best combination can be determined using, for example, a local search method or the like. solution. In addition, in the case where the target position is determined for the XY stage, and the method of scanning by an electric scanner and processing by laser irradiation are used between the movements toward the target position, in fact, it has accuracy or tracking according to the mechanical characteristics of the XY stage And so on. Therefore, as in the third embodiment, it is conceivable that one side of the XY stage is stopped during processing, and only one side of the X direction or the Y direction is realized by stopping the machining using a stage that determines the target position of the XY stage. That is, use the already described processing rectangle division, as shown in Figure 79 division process. That is, the processing rectangle division of the opening position is determined in step 1400, and the order of opening position access in the processing rectangle is optimized in step 1 4 1 0. The operation model of the XY stage is performed in step 1 4 20 to perform the opening in the processing rectangle. Group the hole positions, and set the XY stage control target position of each group. The foregoing step 1400 is the same as step 100 in FIG. 2. -68- (65) 200302761 Step 1410 can be used to solve the initial traveling route setting explained in Fig. 44 and the traveling salesman problem with restrictions where branches at the start and end must be cut. Further, step 1420 can be optimized using the XY stage operation model similar to step 1310 in FIG. 73. In this way, the XY stage operation in one direction of X or Y is performed, and the machining in the next row is processed after the processing of the one row is completed, so that the stage non-stop processing with high positioning accuracy can be realized. Furthermore, in the xy stage operation in one direction of X or γ, the machining rectangle is divided at the position where the hole position exists, so that the non-stop machining can be realized. Furthermore, in the XY stage operation in one direction of X and Y, by performing both X and Y machining on the rectangular division, and selecting the better one, effective non-stop machining can be achieved. In addition, in the XY stage movements in one direction of X and Y, by performing both X and Y processing on the rectangular division, the smaller XY stage movement distance is selected within the resulting rectangular division, and the XY stage is moved. The minimum distance processing rectangle division can realize effective non-stop processing. Moreover, the optimal machining rectangle division in the X or Y axis direction can only limit the operation position of the XY stage to the position with the opening position, so the machining is realized. Speed up. Moreover, by considering the sparse processing rectangle division of the opening position, the position is in the sparse position, and the dense position is in the dense position. The speed advancement method is based on the actual work only, and the work can be performed by 0 5 and 5 lines -69- (66) (66) 200302761 processing, which can realize high-speed processing. In addition, in the positioning performed by both the driving of the XY stage and the scanning of the electric scanner, the opening position is distinguished in the XY stage operation direction by an appropriate width group, and in each group, it is classified in a direction perpendicular to the XY stage. The direction can be processed in a less wasteful path by connecting in a meandering shape. In addition, in the positioning performed by both the driving of the XY stage and the scanning by the electric scanner, by setting the appropriate width in accordance with the density of the opening position of the processing rectangle, a more wasteless path processing can be realized. Furthermore, in the positioning performed by both the driving of the XY stage and the scanning of the electric scanner, the operation time of the electric scanner is modeled in advance so that the sum of the operation time of the electric scanner on the model is minimized and repeated. Improve the access sequence of the hole position, and realize processing with shorter operation time. Furthermore, in the positioning performed by driving in a certain direction of the XY stage and scanning by the electric scanner, the control target position of the XY stage is sequentially set, and in the case of a form such as irradiation being driven toward the target position, By planning the target position in advance, it is possible to efficiently perform non-stop machining. In the aforementioned form, the processing of grouping the positions of the openings from the first visitor to the rectangles of the same size as the scanning area in sequence is performed sequentially, and a good target position can be set by determining the target position. Regarding the aforementioned target position, by making all the points representing the group belonging to the opening position group contained in the rectangular area of the rectangular center position of the scanning area as candidates for the target position, it is possible to increase the flexibility to reduce the driving amount of the XY stage. In the process of optimizing the path throughout the aforementioned target position, a good target position can be determined by making -70- (67) (67) 200302761 the next target position to be the position closest to the previous target position. Or 'In the process of optimizing the path throughout the aforementioned target position, by deciding the next target position temporarily when determining the next target position, the position closest to the target position is minimized from the previous target position. The position where the sum of the cost of moving and the cost of moving to the next target position is the target position can determine the best target position. Or 'in the process of optimizing the path throughout the aforementioned target position, determine several candidates for each target position, and define a combination optimization problem that seeks a combination of candidate position candidates to minimize the sum of the total cost of movement, A good target position can be determined by techniques such as local search. In addition, in the positioning performed by both the driving of the XY stage and the scanning of the electric scanner, the speed and acceleration model of the XY stage are set in advance and driven according to the settings, and the irradiation form is driven during the driving. It is better to plan the speed and acceleration model of the XY stage in advance, so that the non-stop processing of the stage can be effectively performed. Alternatively, in the case of positioning by both the driving of the XY stage and the scanning of the electric scanner, the speed and acceleration model of the XY stage are set in advance, and driven according to the setting, and the irradiation form is driven during the driving. It is better to plan the constant-speed operation speed of the XY stage in advance, which can reduce the processing planning time, and can efficiently perform non-stop machining of the stage. In the aforementioned form, the processing speed can be increased by determining the operating speed for each processing rectangle that divides the hole position. In the aforementioned form, in the execution form of operating the XY stage at a constant speed during image drilling processing -71-(68) 200302761, the earliest irradiation start time in the constant speed operation is determined on the model when planning, and At the latest time of completion of irradiation, if the method of accommodating the maximum speed of the irradiation start time and the time of completion of irradiation is found in this timing (Timing), the maximum limit of constant speed operation can be determined. The aforementioned maximum speed is determined by firstly determining the feasible speed and the non-executable speed. The speed of the interval is investigated for the possibility of implementation. If it is possible, the speed can be updated one by one. If it is not, the speed cannot be updated one by one. The maximum limit of speed. In the aforementioned form, in the execution form that can make the XY stage variable during the drilling process, first, the maximum speed of the constant-speed operation is first determined, and the dense program that restricts this operation is specified. It can make the operation of XY stage faster than constant speed operation. In the aforementioned form, in an execution form that enables the XY stage to be operated in a variable manner, such as in a hole-cutting process, the dense program and the dense program are cut from among the programs that perform the constant speed operation, and the rare program is multiplexed. The process of speeding up the operation speed can speed up the operation of the XY stage. In the aforementioned simultaneous optimization, the function of investigating the possibility of execution is made by accessing the order of a certain hole position and the speed of an XY station. If a certain speed is feasible, the upper limit of the possible speed is increased, if it is not feasible Then improve the order of access to the hole position, make it possible to explore the feasible point access sequence and χγ station speed, if it is feasible, increase the upper limit of the feasible speed, if it is not feasible, lower the lower limit of the impossible speed , Can effectively find the maximum speed of constant speed action. Or, in the aforementioned simultaneous optimization, use the function of -72- (69) (69) 200302761 order to access a certain hole position and the possibility of speed survey implementation of a certain XY station to make the order of maximum access to a certain hole position As a function of the constant speed action speed of the X-axis platform, if the action speed in other access sequences is faster than the action speed in a certain hole position access sequence, the process of repeatedly updating the feasible action speed and point access sequence is repeated as much as possible. , Can effectively find the maximum speed of constant speed action. In addition, regarding the processing rectangle existing at the position of the hole when the irradiation processing is performed while operating the X-axis table, the order of accessing the plurality of processing rectangles and the direction (moving path) for operating the processing rectangles can be optimally determined. Increased processing speed. The aforementioned determination method can increase the processing speed by setting the operation path of the X-axis table to meandering from the end of the substrate if the operation direction of the processing rectangle is made uniform from side to side or up and down. Alternatively, the aforementioned decision method firstly sets the center position of each processing rectangle to the coordinates of each city, applies the traveling salesman problem or the Hamiltonian path minimization problem to determine the access order of the processing rectangles, and secondly determines the action on each processing rectangle. In the direction, the method of determining the entrance of the next processing rectangle of the attention processing rectangle by determining the moving cost of the exit of the previous processing rectangle within two candidates of the entrance and exit repeatedly can determine the optimal χγ stage. Action path. Furthermore, the next processing rectangle of the attention processing rectangle is repeatedly selected as a temporary entrance by making the candidate of the entrance and exit of the next processing rectangle next to the attention processing rectangle less than the temporary entrance. The candidate of the entrance and exit within two can be determined by the sum of the entrance cost of the attention rectangle and the next one. -73- (70) (70) 200302761 The moving cost of the entrance candidate and the moving cost of the next exit candidate and the temporary entrance can be determined. The motion path of the XY stage. In addition, there are two entry candidates for each processing rectangle. When the number of processing rectangles is small, the method of selecting the smallest one by taking the sum of the moving costs for all possibilities can determine the optimal entry and exit selection. In addition, there are two entry candidates for each processed rectangle, and the selection of an excellent entry and exit can be determined by a method such as a local search method. The aforementioned determination method first determines the movement direction of the XY stage of each processing rectangle. Based on the entrance and exit of these processing rectangles, the optimal movement path of the XY stage can be determined according to the solution of the asymmetric salesman problem or the Hamilton path length minimization problem. The aforementioned decision law captures the entrance and exit of each processing rectangle as two cities. It is best to use the solution of the symmetric traveling salesman problem or the Hamilton path length minimization problem based on the restriction that the branches connecting the entrance and exit must be removed. You can determine the order of accessing the processing rectangle and the direction in which the processing area moves on the processing rectangle. Furthermore, when performing non-stop machining, firstly define the opening position access path as a symmetrical traveling salesman problem or a Hamilton path length minimization problem in which each opening position of the entire substrate is a city, and then determine what can be implemented. The target position of the XY stage can realize non-stop machining of the stage. In addition, the opening position of the entire divided substrate is such that one side (progressive edge) is equal to or smaller than the width of the processing area, and the other edge (stop edge) is equal to or greater than the width of the processing area and is parallel to the XY axis. In the middle, the simple movement of the XY stage in the same direction as the X direction or the Y direction is -74- (71) (71) 200302761. If the processing is completed, it moves to the next processing rectangle at high speed, and the speed can be efficiently performed. The plan of the acceleration model. In particular, in the case where each processing rectangle in the above-mentioned processing rectangle division is determined as shown in FIG. 6 or FIG. 10, and the XY stage operation path is set in a meandering shape, a good path with less operation in a non-processing case can be obtained. Moreover, in each of the processing rectangles obtained by dividing by the above processing rectangles, the position where there is no opening is as shown in FIG. 8 and there is a long distance along the progressing side. It can be recognized that the position where the hole position does not exist is the part that extends across the processing rectangle, which makes the XY stage operation speed planning easy. Moreover, by the aforementioned processing rectangle division, as shown in FIG. 10, the directions of the proceeding edge and the stop edge of the full processing rectangle are unified, and the position of the processing rectangle is as shown in FIG. I (B). It is easy to optimize the motion path of the stage, it is possible to move the XY stage with high accuracy, and it is possible to perform a practical XY stage operation speed planning. That is to say, the uniform configuration of the foregoing segments uses the solution of the shortest path problem to minimize the number of segments, and an efficient rectangular configuration can be performed. In addition, in the above-mentioned uniform arrangement, the position of each processing rectangle is finely adjusted by bringing the position of the hole in the processing rectangle close to the center of the processing area, so that high-precision processing can be achieved. Moreover, the aforementioned uniform arrangement of the segments is to minimize the distance between the positions of the openings along a certain direction of the segments within each processing rectangle by selecting the solution that minimizes the number of optimized result segments, The arrangement of the hole positions in each processing rectangle can be effective for high-speed processing. -75- (72) (72) 200302761 In addition, as shown in Fig. 6 or Fig. 10, the direction of the processing rectangle and the stopping edge are unified. When planning the division of the processing rectangle, the two directions are X and Y. Trial, choose the best one, can shorten the processing time. The selection criterion of the aforementioned better one is to select, for example, a small sum of the lengths of the sides of the processing rectangle to obtain a wasteful X γ stage path. Alternatively, by investigating the uniformity of the density of each processing rectangle, the above-mentioned selection criterion can make the operation speed of the XY stage approximately constant for each processing rectangle, and realize wasteless processing. Furthermore, in order to investigate the uniformity of the density of each of the processing rectangles described above, by making cumulative frequency distribution along the edges of the processing rectangles, it is possible to easily and accurately detect the density of the denseness as seen in Figure 15 (B). Both. In addition, when planning the division of the processing rectangle as described in FIG. 6 or FIG. 10 by unifying the direction of the progress and stop sides of the processing rectangle, as shown in FIG. 21, trial is performed for both the X direction and the Y direction. Once the processing rectangle is determined, After the position is determined, the arrangement of the processing rectangle is determined by a good combination of these processing rectangles, which can shorten the processing time. The criterion for the aforementioned good combination is that, for example, by selecting a combination in which the sum of the lengths of the sides of the processing rectangle is small, a wasteful χγ stage path can be obtained. Alternatively, the criterion for the aforementioned good combination is to select the density of each processing rectangle to be a uniform combination, so that the operation speed of the XY table is approximately constant for each processing rectangle, and processing without waste can be realized. In addition, the foregoing processing rectangle segmentation determines whether there is sparseness of the opening positions, and the sparse positions and dense positions exist to distinguish the pore positions according to the degree of sparseness. Regarding the opening positions of the groups other than the most sparse group, Regular -76- (73) (73) 200302761 By determining the location of the processing rectangle first, as shown in Figure 27, the density of the hole positions during processing can be made uniform, which can improve processing efficiency. In addition, in order to determine the density of the above-mentioned hole positions, the entire substrate is divided into a grid shape, and the number of the hole positions of each cell is obtained, and a degree chart is prepared to easily and accurately detect the hole positions. dense. Furthermore, when detecting the regularity of each group of the above-mentioned opening positions, the entire divided substrate is grid-shaped, and the density of the opening positions of each cell is investigated. According to this information, as shown in FIG. A ring of sparse cells surrounds the entire opening position by covering the entire opening position with a cell set rectangle including any cell group with dense opening positions, and regularity can be easily and accurately detected. Moreover, the foregoing processing rectangle division determination is based on the distribution status of the opening positions or the book format of the data file. Whether the hole positions can be divided into the majority group and the minority group as shown in FIG. 31. For the case of bisectability, as shown in FIG. By determining the positions of the processing rectangles for the positions of the holes in the majority group, it is easy to realize the processing rectangle configuration that is consistent in the segment due to the existence of a small number of opening positions, which can improve the processing efficiency. In addition, in order to detect the regularity of the majority / minority of the above-mentioned opening positions, the entire divided substrate is grid-shaped, and the presence or absence of the opening position of each cell is investigated. According to this information, as shown in FIG. 26, A cell set rectangle covering any cell group including the cell at the hole position is covered by a ring of cells at the hole position to cover the full hole position, and regularity can be easily and accurately detected. Furthermore, regarding the obtained cell set For each of the rectangles, by examining the outermost point on the outer side of the rectangle on the outer side of the rectangle, it is possible to more accurately specify the expansion of the opening position described in Pre-77- (74) (74) 200302761, and it is easy to detect regular Configuration. In addition, the regularity of the groups other than the most sparse group and the majority group can be arranged in a straight line or a grid, thereby realizing the scanning rectangle division that can operate the XY stage in the same direction. Furthermore, a linear arrangement is detected from the plurality of cell set rectangles of the group other than the most sparse group and the plurality of cell set rectangles of the majority group, and the processing rectangle representing the processing position is arranged to cover the cell set rectangle, as shown in FIG. 37 ( As shown in A), one or a plurality of divided rectangles are arranged in a straight line with the extending direction of the straight line as the edge. If the interval of the cell set rectangles arranged in a straight line is large to some extent, as shown in FIG. 37 (B) As shown in the figure, by halving the position in front of and behind the position, it can be identified that the position where the hole position does not exist is the part that extends across the processing rectangle, which makes it easy to plan the operation speed of the XY stage. Further, the size of the cell set rectangle arranged in a straight line in a direction perpendicular to the extending direction of the straight line is shown in FIG. 37 (C). 'If it is larger than the size of the processing area, it is arranged in a plurality of straight lines. Machining rectangles can easily obtain machining rectangles. In addition, when covering by a plurality of processing rectangles as described above, 'the inside of the processing rectangle is as shown in FIG. 15 (A)', the situation where the opening position is bisected is shown in Figure 3 7 (D As shown in '), by finely adjusting the center position of the processing rectangle, the arrangement of the opening position in each processing rectangle can be effective for high-speed processing. Furthermore, a grid-like arrangement is detected by a plurality of cell set rectangles of the group other than the most sparse group and a plurality of cell set rectangles of the majority group when the processing rectangle representing the processing position is arranged to cover the cell set rectangle, as shown in FIG. -78- (75) (75) 200302761 As shown, one or a plurality of parallel processing rectangles cover the grid arrangement. Regarding each of the X direction and the Y direction of the grid arrangement, if the interval of the cell collection rectangle is as large as At this level, the rectangular group is divided as shown in Figure 38. At this time, the setting method of the direction of the side when one or a plurality of parallel processing rectangles are arranged so as to cover the cell-set rectangles arranged in a grid pattern is A, the installation of the XY stage (upper stage, lower stage), and By setting the direction to be the upper stage, it is possible to obtain the χγ stage motion path in processing with excellent trackability and accuracy of the XY stage. B. It can be set by reducing the number of processed rectangles, which can reduce the number of times of processing between the processed rectangles. C. By setting the smaller of the sum of the lengths of the sides or the smaller of the sum of the distances between the cell collection rectangles, the time during which the hole position does not appear during processing can be reduced. D. By modeling the action of the electric scanner or the XY stage, the total processing time is actually calculated, and the best one is selected to obtain an effective processing rectangle. Moreover, the foregoing processing rectangle segmentation first determines the sparseness of the opening positions. For the cases where there are sparse positions and dense positions, the hole positions are distinguished according to the degree of sparseness. The regular position determines the processing rectangle position, and the remaining position is left. Next, it is determined whether the hole position can be divided into the majority group and the minority group according to the distribution status of the opening position or the book format of the data file. In the case of the opening positions of most groups, the positions of the processing rectangles are determined by the positions with regularity, the remaining positions are left, and the processing rectangle positions of the remaining positions are finally determined to make the opening positions dense and consistent. The ideal processing rectangle division which can add -79- (76) (76) 200302761 pattern shape without waste can make processing efficiency. In addition, the timing of the machining rectangle positions at the remaining positions is determined. By adopting the arrangement of the machining rectangles with the same segments as shown in Figs. 6, 10, 17, and 21, the machining efficiency can be improved. In addition, when the machining rectangle position of the remaining positions is timing, and there are few positions of the remaining opening positions, processing can be stopped by the application table to improve the processing efficiency. In addition, the use of the stage stop processing when performing the stage stop processing can be captured by the stage stop processing as a part of the stage stop processing, and can be realized by the same control system, making design easy. In addition, in the foregoing embodiment, although the high-speed positioning means is an electric scanner, and the low-speed positioning means is an XY stage, the type or combination of the positioning means is not limited to this, for example, as disclosed by the applicant in Japanese Patent Application Laid-Open No. 2000 0- 7 1 0 8 9 or Japanese Patent Laying-Open 2 0 0 0-3 3 4 6 3 The hybrid processing system (so-called screen cut system) or fast A flash cutting system can also be used. Obviously, the applicable object is not limited to the laser perforating machine that performs point processing. The laser cutting machine that performs linear processing and uses Japanese Patent Application Laid-Open No. 11-149317. A two-head laser processing machine or a general processing machine (for example, a hole-opening device using a mechanical drill) other than a laser beam processing method carried in the storehouse S can be positioned at high speed by driving two or more positioning devices at the same time. All of them are equally applicable. -80-(77) (77) 200302761 [Industrial Applicability] According to the present invention, it is possible to have a plan for effective processing in the preliminary stage of processing, and a low-speed positioning method with high positioning accuracy can be achieved. Excellent effect of shortening processing time by stopping processing. [Brief Description of the Drawings] Fig. 1 is a front view showing the configuration of main parts of a laser hole punch as an example of an object of application of the present invention. Fig. 2 is a flowchart showing a processing procedure according to the first embodiment of the processing plan of the present invention. Fig. 3 is a plan view showing an idea optimized by processing a rectangular division according to the present invention. Fig. 4 is a plan view similarly showing the order of access to the hole positions in the processing rectangle and the optimization of the operation speed of the XY stage. Fig. 5 is a conceptual diagram for optimizing a rectangular motion path in the same manner. FIG. 6 is a schematic diagram of an example of rectangular division processing. Fig. 7 is a plan view showing an example of a substrate in which optimal division is difficult; FIG. 8 is a conceptual diagram of a method of dividing one processing rectangle into a plurality of similarly. Fig. 9 is a conceptual diagram for explaining the minimum distance required when dividing and processing a rectangle. FIG. 10 is a schematic diagram of another example of rectangular division processing. I11 is a plan view showing an example of an arrangement in which segments are identical. -81-(78) (78) 200302761 Fig. 12 is a flowchart showing the procedure of processing the rectangular division in the same arrangement. Fig. 13 is a regularized conceptual diagram showing the processing of rectangular divisions in the same arrangement. Fig. 14 is a line chart showing the regularization concept of the processing rectangle division similarly arranged in the same segment. Fig. 15 is a plan view similarly showing an example of ineffective processing for rectangular division. Fig. 16 is a flowchart showing the sequence of processing plans in the case of processing rectangular divisions in a rectangular arrangement with identical segments. Fig. 17 is a plan view similarly showing the idea of selecting the best method after segmentation by priority in the vertical direction and priority in the horizontal direction. FIG. 18 is a plan view showing an example of the relationship between the density of the hole positions in the processing rectangle and the operating speed of the XY stage. Fig. 19 is a flowchart similarly showing the procedure of selecting the division direction by the uniformity of the density of the processing rectangle. Fig. 20 is an explanatory diagram of a cumulative frequency diagram for investigating the uniformity of the density of processed rectangles in the same manner. Fig. 21 is a plan view similarly showing an example of processing rectangle division by a combination of vertical priority and horizontal priority. FIG. 22 is a plan view showing an arrangement example of opening positions composed of a plurality of pattern shapes (most) and a few other opening positions in the same manner. Fig. 23 is a flow chart showing the process of processing rectangular segmentation at the position of the hole. Fig. 0-82- (79) (79) 200302761 Fig. 24 is a flowchart showing the processing of arrangement of processing rectangles in the dense positions in the aforementioned processing of rectangular segmentation. Fig. 25 is a flowchart showing a processing rectangle arrangement process in the same manner at most positions. Fig. 26 is an explanatory diagram of machining rectangularization by the presence or absence of a point at which a hole is located. Fig. 27 is a flowchart showing a grid division and a grid recognition process in the process of Fig. 24; FIG. 28 is a conceptual diagram of the sparse / dense hierarchical division of the opening position. Fig. 29 is a plan view showing an example of the distribution of dense grids and sparse grids at the opening positions. FIG. 30 is a flowchart showing a grid division and a grid recognition process in the process of FIG. 25. Fig. 31 is a conceptual diagram of the majority / minority group division by the opening position of the present invention. Fig. 32 is a flowchart showing a grid group rectangularization process in the processes of Figs. 24 and 25; Fig. 33 is a flowchart showing a fine adjustment process of the rectangular position of the grid group in the process of Fig. 24; Fig. 34 is a flowchart showing a fine adjustment process of the rectangular position of the grid group in the process of Fig. 25; Fig. 35 is a flowchart showing a process of covering (arrangement of processing rectangles) a rectangular processing group by the grid group in the processing of Figs. 24 and 25; FIG. 36 is an explanatory diagram of the identification method by the arrangement of the regular arrangement of the present invention. (83) (80) (80) 200302761. ® 37 is an explanatory diagram similarly to a coating arranged in a straight line. 38 is an explanatory diagram similarly to the coating arranged in a grid pattern. ® 39 is a flowchart showing an example of the order of opening position access and the order in which XY stage # ^ action is maximized. ® 40 is a flowchart that similarly shows other examples. ® 4 1 is a flowchart showing the other examples as well. ® 42 is a flowchart showing the other examples as well. ® 43 is a flowchart that similarly shows an example in which the order of opening position access is determined in the order of maximizing M operations such as XY stage. FIG. 44 is a flowchart showing a specific example of FIG. 43. FIG. 45 is an explanatory diagram showing an initial path for accessing a hole position. Fig. 46 is a line chart showing an example of the relationship between the division width and the average amount of movement. Fig. 47 is a flowchart showing the procedure of calculating the maximum speed that can be performed for a certain hole position access sequence in the same manner. Fig. 48 is a conceptual diagram of a method for obtaining the maximum speed of the constant velocity of the X-axis table in the same way as the binary method. FIG. 49 is an explanatory diagram of the possibility of execution in the same manner. Fig. 50 is an explanatory diagram of a calculation example in which the earliest start and the latest finish time are similarly obtained. FIG. 51 is an explanatory diagram showing an example of a model that similarly shows the time taken for movement between two points. Fig. 52 is a flowchart of the procedure for obtaining the speed of the low-speed X-axis table in the same manner. -84-(81) (81) 200302761. Fig. 53 is a flowchart of a procedure for obtaining the speed of the high-speed XY stage in the same manner. Fig. 54 is a conceptual diagram of a method for obtaining the maximum speed of the constant velocity of the X-axis table by the proportional method. FIG. 55 is an explanatory diagram of a path update similarly performed by 2 exchanges. FIG. 56 is a conceptual diagram of a path update similarly performed by a nearby Or-opt. FIG. 57 is a conceptual diagram similarly updated by a path near the rear Or-opt. Fig. 58 is a conceptual diagram similarly updated by inserting a nearby path (without inversion). Fig. 59 is a conceptual diagram similarly updated by inserting a nearby path (with inversion). FIG. 60 is a conceptual diagram updated by cross-exchanging two paths in the vicinity. FIG. 61 is a conceptual diagram similarly updated by two paths near 2-opt *. Fig. 62 is a flowchart showing a specific example of a process for optimizing the access sequence of the machining position and a process for maximizing the constant speed of the low-speed positioning means. Fig. 63 is a line diagram showing an example of the operation curve of an XY stage and an electric scanner showing the access sequence of a certain hole position of a certain processing rectangle of a real substrate. -85- (82) (82) 200302761 Figure 64 shows the same speed Variable correspondence thought map. FIG. 65 is a flowchart showing the procedure of a dense program similarly. FIG. 66 is a line chart showing speeds corresponding to variable speeds in the same manner. Fig. 67 is a conceptual diagram of speed multiplexing in the same manner. FIG. 68 is an explanatory diagram of the operation of processing a rectangle. FIG. 69 is an explanatory diagram for optimizing a rectangular motion path in the same manner. Fig. 70 is a flowchart showing the procedure of optimizing the access order of the machining rectangle and the procedure of optimizing the entry and exit of the processing rectangle. Fig. 71 is a flowchart showing the procedure for determining the entrance and exit of the processed rectangle and the optimization of the access sequence. Fig. 72 is a schematic diagram of a simple setting method for processing a rectangular motion path in the same manner. Fig. 73 is a flowchart showing a processing procedure in the second embodiment of the processing plan in the same manner. FIG. 74 is a conceptual diagram of another configuration example. FIG. 75 is an explanatory diagram of a region that is a candidate for the target position in the same manner. FIG. 76 is a conceptual diagram of the target position setting in the XY stage in the same manner. Fig. 77 is an explanatory diagram similarly. Fig. 78 is a diagram similarly. Fig. 79 is a flowchart showing a processing procedure according to the third embodiment of the processing plan of the present invention. [Comparison table of main components] -86-(83) (83) 200302761 1 0: Workpiece 12: XY stage 20: Laser light 22: First electric scanner 23, 25: Rotating mirror 24: Second electric scanner 26 : f- 0 lens