200828138 P53950081TW 22516twf.doc/e 九、發明說明： 【發明所屬之技術領域】 本發明是有關於一種產生糸統與產生方法，特別是有 關於一種決策推論元件的產生糸統與產生方法。 【先前技術】 專豕糸統是早期人工智慧的一個重要分支，且可以視 為一種具有專門知識和經驗的電腦智能程序系統。專家系 ( 統包含使用者介面、推論引擎、知識庫。專家系統是用來 模擬通常由領域專家才能解決的複雜問題。現有的專家系 統通常是針對某一特定用途發展，且内部的現有決策機制 不易修改。因此，現有的專家系統應用的彈性或領域受到 莫大的偈限。 以習知的彈性製造系統（flexible manufaeturing syste% FMS)來說，生產派工(dispatching)的決策(dedsi〇n)是藉由 生^線機台的狀況、加工製程及派工法則等知識所構成的 ， 決策政策，以決定各個加工品分別進行下一製程加工的時 I 間。一旦生產線因機台設備維修、故障、製程變更或派工 規則改變時，習知的彈性製造系統必須費時地改寫程式與 建置系統，•因此無法提供即時便利的修改功能。 【發明内容】 本發明提供一種決策推論元件產生系統，可依圖形化 (graphic)人機介面(human machine interface，HMI)與使用 者互動的結果產生決策推論元件(object)。 本發明&供一種決策推論元件產生方法，可依需求產 200828138 P53950081TW 22516twf.d〇C/e 生決策推論元件，以彈性翻於不_域的系統。 本發明提出-種決策推論元件產 开凡人機介面、:知識庫模組、—連線模組、 板組、-推針擎以及—元件封裝模組。知識庫模 ==圖形化人機介面輪入的知識。連線模組依據使 圖形化人機介面所輸人的設定而連線，以提供與 外界傳輸賴的路徑。決策邏龍 f200828138 P53950081TW 22516twf.doc/e IX. Description of the Invention: [Technical Field of the Invention] The present invention relates to a method for generating a system and a method for generating the same, and more particularly to a method and method for generating a decision inference element. [Prior Art] Specialized system is an important branch of early artificial intelligence and can be regarded as a computer intelligent program system with expertise and experience. Expert system (including user interface, inference engine, knowledge base. Expert system is used to simulate complex problems that are usually solved by domain experts. Existing expert systems are usually developed for a specific purpose, and internal decision mechanisms are internal. It is not easy to modify. Therefore, the flexibility or field of the existing expert system application is greatly limited. In the case of the flexible manufaeturing syste% FMS, the decision of the dispatching (dedsi〇n) It is made up of the knowledge of the condition of the machine, the processing technology and the dispatching rules, and the decision-making policy to determine the time between each processing product for the next process. Once the production line is repaired by the machine equipment, When a failure, a process change, or a dispatch rule is changed, the conventional flexible manufacturing system must time-rewrite the program and the build system, and thus cannot provide an instant and convenient modification function. [Invention] The present invention provides a decision inference component generation system. Can interact with the user according to the graphical human machine interface (HMI) The result produces a decision inference object. The present invention is a method for generating a decision inference component that can be used to produce a 200828138 P53950081TW 22516twf.d〇C/e decision inference component that is flexible to a non-domain system. The invention proposes a decision-making inference component to produce a mortal machine interface, a knowledge base module, a connection module, a board group, a push pin engine, and a component package module. Knowledge base module == graphical human machine interface The knowledge of the turn-in. The connection module is connected according to the settings of the person who inputs the graphical human-machine interface to provide a path to the outside world.
化人機介面編輯的決策邏輯 圖形 輯的、朿_用以構成—決策推論模型。推論引擎經由圖 瓜化人機〃面與使用者在推論·巾互動，峨決策推論 模^推論產生-絲推論結果。元件封裝模組用以將決策 推雨板型及其使用之知識、資訊與決策邏輯封裝為一決策 推論元件。 在本發明之-實施例中，決策推論元件產生系統更包 括一決策推論驗證，經由_化人機介面與使用者在 驗證過程中絲’崎證決策推論模型是否正確。 在本發明之-實施例中’上述之連線模組被設定以從 -外部貧料庫、-可程式化邏輯㈣器㈣gfammable l〇gic C〇ntr〇Uer，PLC)與一射頻標籤（radio frequency identification，RFID)至少其中之一傳輸資訊。 在本發明之-實施例中，連線模組被設定以將決策推 論結果傳輸至一外部資料庫。 在本發明之-實施例中’上述之決策推論模型與決策 推論結果之格式包括可擴展標示語言（extensible markup 200828138 P53950081TW 22516twf.doc/e language, XML) 〇 在本發明之一實施例中，上述之決策邏輯模組以公 式檀木圖片、戒息、電話號碼及電子郵件地址至少其 中之一為規則前提或觸發事件。 在本發明之-實施例中，上述之決策邏輯模組所儲存 的決策邏輯建構為多個規則，達些規則中至少一部分會組 成多個規則群組，每-酬群組⑽這些規則無順序ς、。 其卜規麟組可依順糊發，而這些_群組之觸發順 不勉月冉徒出住的來雅黹兀仵產生方法，其包括： 經由-圖形化人機介面輸人知識並儲存至—知識庫模板；· ，由圖形化人機介面輸人-連線模組的連線設定，以提供 路徑;經由圖形化人機介面編輯決策邏 =的決策邏輯用以構成—決策推論模型;The decision logic of the human-machine interface editing, the __ is used to constitute the decision-making inference model. The inference engine interacts with the user in the inference and the towel through the diagram, and the decision inference is derived from the inference. The component packaging module is used to encapsulate the decision-making push-pull type and its knowledge, information and decision logic into a decision inference component. In the embodiment of the present invention, the decision inference component generation system further includes a decision inference verification, and whether the model is correct by the user interface and the user in the verification process. In the embodiment of the present invention, the above-mentioned connection module is set to be an external poor library, a programmable logic (four) device (four) gfammable l〇gic C〇ntr〇Uer, PLC) and a radio frequency tag (radio) Frequency identification, RFID) at least one of the transmission of information. In an embodiment of the invention, the wiring module is configured to transmit the decision inference results to an external database. In the embodiment of the present invention, the format of the above-described decision inference model and decision inference result includes an extensible markup language (extensible markup 200828138 P53950081TW 22516twf.doc/e language, XML). In an embodiment of the present invention, the above The decision logic module uses at least one of the formula sandalwood picture, the ringing number, the phone number and the email address as a rule premise or a trigger event. In the embodiment of the present invention, the decision logic stored by the decision logic module is constructed as a plurality of rules, and at least some of the rules form a plurality of rule groups, and each rule group (10) has no rules. Oh,. The group of arbitrators can be succinct, and the triggers of these _ groups are not in the way of the 黹兀仵 出 出 出 , , , , , , , , , , , , , 图形 图形 图形 图形 图形 图形 图形 图形 图形 图形 图形 图形 图形 图形 图形 图形To - knowledge base template; ·, by the graphical human-machine interface input-connection module connection setting to provide the path; through the graphical human-machine interface editing decision logic = decision logic to constitute - decision inference model ;
Ci 人2面與—推論引擎互動，峨決隸論模難 一決桌推論絲；以將觀策 識、^與決策邏輯-併輯為—決策推論=使用之知 ，本發明之-實施例中，決策推論元件 ▲，決策推論模型及其使用之: -決二上 圖形化人機介面與 括4r—，=二=更確包 括經由連線模組而從-外部資料庫、—可程式化邏輯= 200828138 P53950081TW 22516twf.doc/e 器與一射頻標籤至少其中之—傳輸資訊。 咏在本發明之一實施例中，更包括經由連線模組而將決 朿推論結果傳輸至一外部資料庫。 在本發明之一實施例中，上述之決策推論模型與決策 推論結果是以可擴展標示語言格式表現。 土在本發明之-實施例中，上述之圖形化人機介面是以 ^端操控模式被操控。 综上所述，在本發明的決策推論元件產生系統與方法 ’因採用了圖形化人機介面，所以使用者可省時且便利 地完成整憾策推論的龍、驗證與推論玉作。此外，使 用者透過連雜組可以將所需的知識、資贿決策規則等 匯入決策推論元件產生系統中，如此可讓決策推論的工作 更^有彈性，也可省下重複建置這些資料的時間。另外， 、策推_元件產生系統所產生的推論元件可以適時地應用 到不同的系統及裝置，且可依環境狀況改變來做調整。 為讓本發明之上述特徵和優點能更明顯易懂，下文特 舉較佳實施例，並配合所附圖式，作詳細說明如下。 【實施方式】 圖1A為本發明一實施例之決策推論元件產生系統應 用於彈性製造系制示意圖，而圖1β為本發明—實施例 之決策推論元件產生方法的流程圖。請參照圖1A，本發明 八只苑例之決策推淪元件產生系統1〇〇包括一圖形化人機 二面110、一知識庫模組120、一連線模組13〇、一決策邏 輯模組H0、一推論引擎150以及一元件封裝模組16〇。知 8 200828138 P53950081TW 22516twf.doc/e 識庫板組120儲存使用者50經由圖形化人機介面11〇輸入 的知識。連線模組130依據使用者50經由圖形化人機介面 110所輸入的設定而連線，以提供與外界傳輸資訊的路 徑，而這些例如由外部設備、感測器或外部資料庫等經由 連線模組130傳入的知識與資訊也可儲存於知識庫模組 120中。決策邏輯模組140儲存使用者50經由圖形化人機 介面110編輯的決策邏輯。 推論引擎150經由圖形化人機介面110與使用者5〇 在推論過程中互動，並以使用者5〇輸入的資訊與知識及編 輯的決策邏輯所構成的決策推論模型予以推論而產生_決 策推論結果。元件封裝模組160用以將該決策推論模型及 其所使用之知識、資訊與決策邏輯一併封裝為一決策推論 元件。 口田 請同時參照圖1A與圖1B，本實施例之決策推論元件 產生方法可使用決策推論元件產生系統1〇〇,其步驟如下 所述。首先，使用者50經由圖形化人機介面11〇輪入知識 並儲存至知識庫模組12〇(步驟S2〇2)，其中知識例如為^ 件的種類、加工的方式、派工法則等各類經驗及知識:、言 些知識可以文字、圖片、公式或其他適當形式呈現。= 進仃步驟S204，使用者50經由圖形化人機介面11〇 連線模組⑽的連線蚊，·設定連_介面並選^ ，入或匯出資訊的裝置等，以提供與外界傳輸資訊的路 徑。接著，使用者50經由圖形化人機介面11〇以例如拖 多種圖示（icon)編輯決策樹的方式或藉由點選選單 9 200828138 P53950081TW 22516twf.doc/e box)、文字方塊(text box)、核取方塊(check box)、選擇I丑 (radio button)來編輯決策邏輯，並將這些決策邏輯儲存至 決策邏輯模組140(步驟S206)。在本實施例中，決策邏輯 是用來分派加工工件加工的順序與流程。 然後進行步驟S208，使用者50經由圖形化人機介面 110與推論引擎150互動，使推論引擎15〇以使用者輸入 的資汛與知識及編輯的決策邏輯所構成的決策推論模型予 以，論而產生一決策推論結果。最後進行步驟S21〇，將該 決策推論模型及其使用之知識、資訊與決策邏輯封裝為一 決策推論元件。在本實施例中，所產生的決策推論元件可 植入彈性製造系統6〇〇，讓彈性製造系統可以依推 為的結果來決定加工的順序及工件的分派。 決策推論元件產生系統100提供了圖形化人機介面 110，使得使用者50可以直接操作或遠端操控的方式，藉 由才也曳圖不、點選選單(list b〇x)、文字方塊咖对b〇x)、勾Ci people's two sides interact with the inference engine, and it is difficult to make a decision on the table; to view the knowledge, ^ and decision logic - and to make a decision - inference = use, the present invention - the embodiment In the decision-making inference component ▲, the decision inference model and its use: - The graphical human-machine interface on the second and the 4r-, = two = more includes the connection module from the external database, - programmable Logic = 200828138 P53950081TW 22516twf.doc/e and at least one of the RF tags - transmitting information. In an embodiment of the invention, the method further comprises transmitting the decision inference result to an external database via the connection module. In one embodiment of the invention, the decision inference model and the decision inference result described above are expressed in an extensible markup language format. In the embodiment of the invention, the graphical human interface described above is manipulated in the ^ terminal control mode. In summary, in the decision making component generation system and method of the present invention, the user can save the time, convenience and convenience of the dragon, verification and inference work. In addition, the user can transfer the required knowledge and bribe decision rules into the decision inference component generation system through the mixed group, so that the work of the decision inference can be made more flexible, and the repeated construction of the data can be saved. time. In addition, the inference components generated by the _ component generation system can be applied to different systems and devices in a timely manner, and can be adjusted according to changes in environmental conditions. The above described features and advantages of the present invention will become more apparent from the following description. [Embodiment] FIG. 1A is a schematic diagram of a decision inference component generation system applied to an elastic manufacturing system according to an embodiment of the present invention, and FIG. 1 is a flowchart of a method for generating a decision inference component according to the present invention. Referring to FIG. 1A, the decision-making push component generation system of the eight examples of the present invention includes a graphical human-machine two-sided 110, a knowledge base module 120, a connection module 13〇, and a decision logic module. The group H0, a inference engine 150, and a component package module 16A. Knowledge 8 200828138 P53950081TW 22516twf.doc/e The library board 120 stores the knowledge that the user 50 inputs via the graphical human interface 11〇. The connection module 130 is connected according to the settings input by the user 50 via the graphical human interface 110 to provide a path for transmitting information to the outside world, such as by an external device, a sensor, or an external database. The knowledge and information introduced by the line module 130 can also be stored in the knowledge base module 120. The decision logic module 140 stores decision logic that the user 50 edits via the graphical human interface 110. The inference engine 150 interacts with the user 5〇 in the inference process via the graphical human-machine interface 110, and infers from the decision inference model formed by the information input by the user 5 and the decision logic of the knowledge and editing. result. The component packaging module 160 is used to encapsulate the decision inference model and the knowledge, information and decision logic used by it into a decision inference component. Mouth Field Referring to FIG. 1A and FIG. 1B simultaneously, the decision inference component generation method of the present embodiment can use the decision inference component generation system 1〇〇, the steps of which are as follows. First, the user 50 enters knowledge into the knowledge base module 12 via the graphical human-machine interface 11 (step S2〇2), wherein the knowledge is, for example, the type of the workpiece, the processing method, the dispatching rule, and the like. Class experience and knowledge: Words can be presented in words, pictures, formulas or other appropriate forms. In step S204, the user 50 provides a connection with the outside world via a graphical human-machine interface 11 connecting the mosquito module of the connection module (10), setting a connection interface, and selecting or importing information. The path to the information. Then, the user 50 via the graphical human interface 11 to edit the decision tree by, for example, dragging multiple icons or by clicking the menu 9 200828138 P53950081TW 22516twf.doc/e box), text box The check box is selected, the radio button is selected to edit the decision logic, and the decision logic is stored in the decision logic module 140 (step S206). In this embodiment, the decision logic is used to assign the order and flow of the machining of the workpiece. Then, in step S208, the user 50 interacts with the inference engine 150 via the graphical human-machine interface 110, so that the inference engine 15 provides the decision inference model composed of the user-entered assets and the decision logic of knowledge and editing. Produce a decision inference result. Finally, in step S21, the decision inference model and the knowledge, information and decision logic used therein are encapsulated into a decision inference component. In this embodiment, the resulting decision inference component can be implanted into the flexible manufacturing system 6〇〇, allowing the flexible manufacturing system to determine the order of processing and the assignment of the workpiece based on the results of the push. The decision inference component generation system 100 provides a graphical human interface 110 so that the user 50 can directly or remotely control the way, by dragging the map, selecting a list (list b〇x), text box coffee For b〇x), tick
取核取方塊(check box)或點取選擇鈕(radi〇 bmt〇n)等方式 ，決策推論元件產生系統來互動，即可省時且便利地 ^整個絲推論建模、驗證與推論的功。在本實施例 丄圖形化人機介面110並非限定必須以圖片方式顯示各 對=方塊，而是相較於—般直接以程式語言撰寫決策推論 二ί方式而言’ _化人機介面11G可提供直覺式的圖 ::面，省去使用者5。必財記程式語言之 指令名稱等負擔。 此外，透過设定連線模、组130，使用者50可以將所需 200828138 P53950081TW 22516twf.doc/e 策輯轉^策減元件產生系統 1〇〇中’如此可讓決策推論的工作更具有彈性，也可 重複建置這些資料的時間。 策推論元件產生系統100所產生的推論元件 可以翁地__如：雜製料統議、物流派送、 人貝排班、生觸程、運紅具航轉程、醫療處方、金 =投資！測與預警、投資組合、文字搜尋擷取辨識、Take the check box or the radi〇bmt〇n method to make the decision inference component generate system to interact, which can save time and convenience ^ the whole silk inference modeling, verification and inference work . In the embodiment, the graphical human-machine interface 110 is not limited to displaying each pair of squares in a picture manner, but rather than directly writing the decision-making inference method in the programming language. Provides an intuitive picture: face, eliminating the user 5. It is necessary to bear the burden of the instruction name of the programming language. In addition, by setting the connection mode and group 130, the user 50 can turn the required 200828138 P53950081TW 22516twf.doc/e strategy into the component generation system. This makes the decision inference more flexible. , the time to re-create these materials can also be repeated. The inferential component produced by the system component generation system 100 can be __ such as: miscellaneous materials, logistics delivery, human shell scheduling, raw contact, red shipping, medical prescription, gold = investment! Measurement and early warning, investment portfolio, text search, identification,
止乂 S運决策、健康養生休閒建議、智慧型機械人、智藝 =車輛、語音辨識、安全警報系統、老人絲遠端照護^ 養預警系統、環境空間搜尋，環境生態監測預警、自 源仏勘或其他需進行決策推論的領域。以彈性製造系統 600來說，生產線因機台設備故障或是改變製料，可以 再次進行步驟S202至步驟S21〇。使用者5〇可以透過圖形 化人機介面11G’很容易將故障的機台自^定的知識庫中 移除’使得決策推論元件產生系統⑽重新產生一決 策推論模型’並運用此決策推論模型封裝產生—新的決/策 推論元件。 在決策推論元件植入彈性製造系統6〇〇後，可使得彈 性製造系統600得以依新產生的決策推論模型予以推論產 生新的決桌推論結果而進行加工的分派。因此，決策推論 元件產生系統100可動態地修改決策的機制以調整製程， 使知彈性製造系統600得以依新修訂的派工規則進行重新 派工而快速地恢復生產的工作。 圖2A為本發明另一實施例之決策推論元件產生系統 11 200828138 P53950081TW 22516twf.doc/e 應用於彈性製造系統的示意圖，而圖2B為本發明另一奋 施例之決策推論元件產生方法的流程圖。圖2八之決策推 論元件產生系統102與圖ία之決策推論元件產生系統 相似，而圖2B之決策推論元件產生方法則與圖m '之決策 推論元件產生方法相似，其中相同元件或步驟使用相同標 號並省略其介紹。請參_2Α，本實關之絲推論元; 產生系統102更包括一決策推論驗證模組17〇。請參考圖 2Β ’在由使用者輸人的資訊與知識及賴的決策邏輯所構 成決策推論模型後(步驟S208)，決策推論驗證模組17〇經 由圖形化人機介面110與使用者5〇在驗證過程中互動，= 驗證決策推論模型是否正確(步驟S214)。 此外，請對照參考圖2A與圖2B，連線模組13〇可設 定與外部資訊60作賴，峨外部#料庫a、可程式化 邏輯控制器64與射頻標籤66至少其中之一傳輸資訊(步驟 S212)。另外，在設定完連線模組且驗證完決策推論模型是 否正確之後(步驟S214)’使用者50還可將使用決策推論模 ^所推論產生的-決策推論結果經由連線模組傳輸至外部 貝料庫92或是其他的外部資料庫（未繪示）（步驟S2i6)。 另外，在本實施例中，決策推論模型及決策推論結果 之格式可以是可擴展標示語言，使得決策推論模型可以相 谷於適合的轉系統或與其他系統整合。力策邏輯模組 可以a式、檔案、圖片、汛息、電話號碼及電子郵件 地址至少其巾之—為酬前提_發事件。決策邏輯模組 14〇所儲存的決策邏輯可建構為多個規則，這些規則中至 12 200828138 P53950081TW 22516twf.doc/e 少一部分的規則可分別組成多個規則群組，而每一個規則 群組内的規則並不一定要有順序性。這些規則群組可選擇 性地依順序觸發，且這些規則群組之觸發順序也可調整。 為了更詳盡說明決策推論元件產生系統1〇2應用於彈 性製造系統600的情形，以下以彈性製造系統6〇〇為主來 做#明。圖3A為應用圖2A之決策推論元件產生系統的彈 性生產糸統的示意圖。請參照圖3 A，彈性製造系統600 包括生產線600a以及控制系統6〇〇b。生產線600a包括一 自動存取糸統(automated storage / retrieval system)610、一 運送軌道620以及多個電腦數值控制(c〇mpUter numericai control，CNC)加工機630。自動存取系統610包括多個運 送盤(pallet)610a〜610i。其中，運送盤6術〜6101藉由執道 式搬運車(rail guided vehicles，RGV)(未繪示)在運送轨道 620上移動，使得位於運送盤6i〇a〜61〇i上的工件可以運 送至電腦數值控制加工機630進行加工。 請一併參考圖3A、圖2A與圖2B，使用者50經由圖 形化人機介面11〇在決策推論元件產生系統1〇2中選取工 件的種類、加工的方式、派工法則（步驟S2〇2)。使用者5〇 進行連線的設定，設定欲透過連線模組13〇存取的外部資 料庫、工件的射頻標籤等(步驟S204)。連線模組130依使 用者no選取的結果自外部資料庫取出資訊(步驟S212)。 使用者50經由圖形化人機介面n〇建構出如圖3B的決策 邏輯(步驟S206)。 接著’推論引擎150以使用者輸入的資訊與知識及編 13 200828138 P53950081TW 22516twf.doc/e 輯的決策邏輯所構成的決策推論模型開始推論整個加工流 私，依各個工件加工的順序作一個整體的時間規劃，讓彈 性製造系統600能流暢地運送工件而減少各工件彼此等待 的時間，最後推論產生一決策推論結果(步驟S2〇8)，並可 經由圖型化人機介面與一決策推論驗證模組互動，以驗證 決策推論模型是否正確(步驟S214)。 圖3C為目3A的彈性生產系統在另一時間點的示意 ( 圖。舉例來說，請一併參考圖3A〜圖3C，運送盤61〇c上 的工件完成一加工的作業後，等待要運送至電腦數值控制 加工機630b(步驟S702)，以進行下一階段的加工作業。步 驟S704選擇先進先出的派工規則。假設運送盤61〇b上的 工件尚未完成在電腦數值控制加工機630b加工的作業，且 佔據了暫存區(步驟S708)。此時，由於運送盤61〇c還在 電腦數值控制加工機630c上，因此運送盤61〇()需送回自 動存取系統(步驟S722)。最後進行步驟S724，把分派的結 果寫入外部資料庫中，以進行下一個派工過程。 ( 請一併參考圖3A、圖2A與圖2B，在推論的結果產 生(步驟S208)後，使用者50再經由圖形化人機介面11〇 ，決策推論驗證模組17〇互動，確認推論的結果是否有衝 大，是是否按照排定的順序進行作業(步驟S214)。在確認 ^策推論模型無誤後，則讓元件封裝模組160將決策推論 核型、該彈性製造系統知識庫、決策邏輯與派工規則封裝 為二決策推論元件(步驟S2l〇)，以植入控制系統6〇%中， 使得控制系統600b得以自動地掃描該彈性製造系統生產 200828138 P53950081TW 22516twf.doc/e 線實況來進娜論，並依縣―:欠決策推論的結果來控 生產線600a。 綜上所述’本發明決策推論元件產生系統與決策推論 兀件產生方法具有下列優點： -、決策推論元件產生系統提供了圖形化人機介面， 使得使用者可斜且便概完歧綠論的建置工作。 之 二、使用者透過對連線模組作設定可以將其所使用 知識、資訊與決策邏輯等匯人決策推論元件產生系統中， 枝推論的工作更具有雜，也可省下重複建置 这些貧料的時間。 μ ί、決策推論元件產生純所產生的推論元件可以適 U用到不同㈣統及裝置’且可依環境狀況改變 速且彈性地做出對應的調整。 發明已啸佳實施觸露如上，然其並非用以 ’任何所屬技術領域巾具有通常知識者，在不 因此之精神和範圍内，當可作些許之更動與潤飾， =本發明之健簡當視_之申請專魏_界定者 【圖式簡單說明】 田^认為本發明—實關之決策推論元件產生系统庫 用於彈性製衫統料賴。 玍糸、，充應 流程Ξm為本發明—實施歉決策推論元件產生方法的 圖2a為本發明另一實施例之決策推論元件產生系統 15 200828138 P53950081TW 22516twf.doc/e 應用於彈性製造系統的示意圖。 圖2B為本發明另一實施例之決策推論元件產生方法 的流程圖。 圖3A為應用2A之決策推論元件產生系統的 生產系統的示意圖。 意圖 圖3B為圖3A之彈性生產系統所使用之決策邏輯的示Stop S transport decision, health health leisure advice, intelligent robot, intelligence = vehicle, voice recognition, security alarm system, remote care for the elderly, nursing early warning system, environmental space search, environmental ecological monitoring and warning, self-sourced Survey or other areas where decision making is required. In the case of the elastic manufacturing system 600, the production line may perform step S202 to step S21 again due to failure of the machine equipment or change of the material. The user can easily remove the faulty machine from the fixed knowledge base through the graphical human-machine interface 11G', so that the decision inference component generation system (10) regenerates a decision inference model and uses the decision inference model. Encapsulation produces - new decision/inference components. After the decision inference component is implanted into the elastic manufacturing system, the elastic manufacturing system 600 can be inferred from the newly generated decision inference model to generate a new table inference result for processing. Thus, the decision inference component generation system 100 can dynamically modify the decision making mechanism to adjust the process so that the flexible manufacturing system 600 can be reworked in accordance with the newly revised dispatch rules to quickly resume production. 2A is a schematic diagram of a decision inference component generation system 11 200828138 P53950081TW 22516twf.doc/e applied to an elastic manufacturing system, and FIG. 2B is a flow of a method for generating a decision inference component according to another embodiment of the present invention; Figure. The decision inference component generation system 102 of FIG. 2 is similar to the decision inference component generation system of FIG. 2A, and the decision inference component generation method of FIG. 2B is similar to the decision inference component generation method of FIG. 4', in which the same component or step uses the same The label is omitted and its description is omitted. Please refer to _2 Α, the actual quotation of the wire; the production system 102 further includes a decision inference verification module 17 〇. Referring to FIG. 2Β 'after the decision inference model formed by the information and knowledge input by the user and the decision logic of the user (step S208), the decision inference verification module 17 communicates with the user via the graphical human interface 110. Interacting during the verification process, = verifying that the decision inference model is correct (step S214). In addition, referring to FIG. 2A and FIG. 2B, the connection module 13 can be configured to communicate with the external information 60, and at least one of the external #库库, the programmable logic controller 64, and the radio frequency tag 66 transmits information. (Step S212). In addition, after setting the connection module and verifying that the decision inference model is correct (step S214), the user 50 can also transmit the result of the decision inference derived from the inference using the decision inference module to the outside via the connection module. The billet library 92 or other external database (not shown) (step S2i6). In addition, in the present embodiment, the format of the decision inference model and the decision inference result may be an extensible markup language, so that the decision inference model can be integrated with a suitable transfer system or integrated with other systems. The tactical logic module can be used for a type, file, picture, suffocation, phone number and e-mail address. The decision logic stored by the decision logic module 14 can be constructed into a plurality of rules, and among these rules, a part of the rules can be composed of a plurality of rule groups, and each rule group is The rules do not have to be sequential. These rule groups are optionally triggered in sequence, and the order in which these rule groups are triggered can also be adjusted. In order to explain in more detail the case where the decision inference component generation system 1〇2 is applied to the elastic manufacturing system 600, the following is based on the elastic manufacturing system. Figure 3A is a schematic illustration of an elastic production system employing the decision inference component generation system of Figure 2A. Referring to FIG. 3A, the elastic manufacturing system 600 includes a production line 600a and a control system 6〇〇b. The production line 600a includes an automated storage/review system 610, a transport track 620, and a plurality of computer numerical control (CC) machines 630. The automatic access system 610 includes a plurality of transport pallets 610a through 610i. Wherein, the transport tray 6~6101 is moved on the transport rail 620 by means of a rail guided vehicle (RGV) (not shown), so that the workpieces on the transport trays 6i〇a~61〇i can be transported. The processing is performed to the computer numerical control processing machine 630. Referring to FIG. 3A, FIG. 2A and FIG. 2B together, the user 50 selects the type of the workpiece, the processing method, and the dispatching rule in the decision inference component generating system 1〇2 via the graphical human-machine interface 11 (step S2〇). 2). The user 5 sets the connection, sets the external material library to be accessed through the connection module 13 , the radio frequency tag of the workpiece, and the like (step S204). The connection module 130 retrieves information from the external database based on the result selected by the user no (step S212). The user 50 constructs the decision logic of Fig. 3B via the graphical human interface n (step S206). Then, the inference engine 150 begins to infer the entire processing flow with the information and knowledge input by the user and the decision inference model composed by the decision logic of the series, and the overall processing is performed according to the order of processing each workpiece. The time planning allows the elastic manufacturing system 600 to smoothly transport the workpiece and reduce the waiting time of each workpiece. Finally, the inference results in a decision inference result (step S2〇8), and can be verified by the graphical human-machine interface and a decision inference. The modules interact to verify whether the decision inference model is correct (step S214). Fig. 3C is a schematic view of the elastic production system of the item 3A at another time point (Fig. 3, for example, referring to FIG. 3A to FIG. 3C, after the workpiece on the transport tray 61〇c completes a processing operation, waiting for It is transported to the computer numerical control processing machine 630b (step S702) to perform the next stage of the machining operation. Step S704 selects the first-in first-out dispatch rule. It is assumed that the workpiece on the transport tray 61〇b has not been completed in the computer numerical control processing machine. 630b processed work, and occupies the temporary storage area (step S708). At this time, since the transport tray 61〇c is still on the computer numerical control processing machine 630c, the transport tray 61〇() needs to be sent back to the automatic access system ( Step S722). Finally, step S724 is performed, and the dispatched result is written into the external database for the next dispatching process. (Please refer to FIG. 3A, FIG. 2A and FIG. 2B together, and the result of the inference is generated (step S208). After that, the user 50 further interacts with the decision inference verification module 17 via the graphical human-machine interface 11 to confirm whether the result of the inference is large or not, and whether the job is performed in the scheduled order (step S214). ^Inference model After the error, the component packaging module 160 encapsulates the decision inference kernel, the elastic manufacturing system knowledge base, the decision logic and the dispatch rule into two decision inference components (step S2l〇) to be implanted into the control system 6〇%. The control system 600b is enabled to automatically scan the flexible manufacturing system to produce the 200828138 P53950081TW 22516twf.doc/e line to enter the narrative, and control the production line 600a according to the results of the county->indecision inference. The decision inference component generation system and the decision inference component generation method have the following advantages: - The decision inference component generation system provides a graphical human-machine interface, so that the user can complete the construction of the disambiguation theory. By setting the connection module, the user can use the knowledge, information and decision logic used in the decision-making inference component generation system. The work of branch inference is more complicated, and the redundant construction can be saved. The time of the μ ί, the decision inference component produces purely generated inferential components that can be used in different (four) systems and devices' and can change speed according to environmental conditions and Responsively make corresponding adjustments. The invention has been implemented as above, but it is not used to 'have any general knowledge of the technical field, and in the spirit and scope of this, when there are some changes and refinements , = the application of the invention, the application of the _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ FIG. 2a is a schematic diagram of a decision inference component generation system according to another embodiment of the present invention. 200828138 P53950081TW 22516twf.doc/e A schematic diagram applied to an elastic manufacturing system. 2B is a flow chart of a method for generating a decision inference component according to another embodiment of the present invention. Figure 3A is a schematic illustration of a production system for a decision inference component generation system employing 2A. Intent Figure 3B is an illustration of the decision logic used by the elastic production system of Figure 3A.
a圖3C為圖3A的彈性生產系統在另一時間點的八立 【主要元件符號說明】 50 :使用者 60 :外部資訊 62 :外部資料庫 64 :可程式化邏輯控制器 66 :射頻標籤 100、102 :決策推論元件產生系統 110 ·•圖形化人機介面 120 ··知識庫模組 130 :連線模組 140 :決策邏輯模組 150 :推論引擎 160 :元件封裝模組 170 :決策推論驗證模組 600 :彈性製造系統 16 200828138 P53950081TW 22516twf.doc/e 600a :生產線 600b ··控制系統 610 ··自動存取系統 610a〜610i :運送盤 620 :運送軌道 630 :電腦數值控制加工機 S202〜S216 :決策推論元件產生方法之步驟 S702〜S724 :決策邏輯的各步驟a Figure 3C shows the elastic production system of Fig. 3A at another point in time [main component symbol description] 50: user 60: external information 62: external database 64: programmable logic controller 66: radio frequency tag 100 102: Decision Inference Component Generation System 110 • Graphical Human Machine Interface 120 • Knowledge Base Module 130: Connection Module 140: Decision Logic Module 150: Inference Engine 160: Component Package Module 170: Decision Inference Verification Module 600: Elastic Manufacturing System 16 200828138 P53950081TW 22516twf.doc/e 600a: Production Line 600b · Control System 610 · Automatic Access System 610a~610i: Transport Disk 620: Transport Track 630: Computer Numerical Control Processing Machine S202~S216 : Steps S702 to S724 of the decision inference component generation method: steps of the decision logic
1717