JP3764233B2 - Abnormality inspection method and apparatus - Google Patents

Description

【００９９】
ここで、ｉは周波数成分を示すもので、ｉ＝１…ｎである。また、ｋは被検差物の数でｋ＝…１Ｎであり、Ｐ_EMkiはｋ番目の被検査体の周波数成分ｉについてのピーク値である。上記式(1)(2)によって求められた平均値Ｘｉ´及び標準偏差σ_i に基づいて各周波数成分ごとにＲＯＭ１８に格納されている次式(3) により設定値Ｐ_T1…Ｐ_Tnを求める。
【０１００】
【数２】

【０１０１】
（ただしｉ＝１…ｎ）
ここで、Ｚは検査環境、検査基準に応じて任意に設定できる変数である。
上記設定値Ｐ_T1…Ｐ_Tn、有効最大ピーク値Ｐ_EM1 …Ｐ_EMn 及び異常の有無についての判定結果は、ステップ＃７において、図８及び図９に示すように各周波数成分ごとにＣＲＴ２８により表示される。
【０１０２】
これらの図において周波数成分の数ｎは２８となっている。なお、ＣＲＴ２８における表示のためのプログラムはＲＯＭ１８に予め格納されている。
図８は、波形が図７に示すような被検査体１の回転部分が固定部分に接触したときのいわゆる「当り異常」に起因するときの検査結果のＣＲＴ表示を示している。
【０１０３】
この図中において、棒グラフの影線部分は異常部分を示していて、有効最大ピーク値が設定値より大きいときに「異常」と判定している。異常を表わしている帯域は異常を示す棒グラフの直上部に矩形３９…により表示されている。
【０１０４】
このように上記第１の実施の形態においては、被検査体１に対して外部の音を第２の音検出器２０１により検出し、この検出された音の振幅が予め設定されたしきい値を越えると、少なくともサンプリング中の被検査体１の音の振幅のデータを削除するので、被検査体１の音の影響を受けることなく、予期せぬ衝撃性の暗騒音や暗振動を確実に検出して、確実に被検査体１の異常判定ができる。
(2) 次に本発明の第２の実施の形態について説明する。なお、図１と同一部分には同一符号を付してその詳しい説明は省略する。
【０１０５】
この第２の実施の形態は、図１に示す異常検査装置を図９に示す被検査体ＳＴ0 に適用したものである。以下、図１に示す異常検査装置に従い説明する。
回転機器等の被検査体ＳＴ10は、ベースＳＴ11上に各防振ゴムＳＴ12を介して載置されている。ベースＳＴ11上には、支柱ＳＴ13が立設され、この支柱ＳＴ13の先端部に防振ゴムＳＴ14を介してシリンダＳＴ15が設けられている。
【０１０６】
そして、このシリンダＳＴ15の駆動先端に防振ゴムＳＴ16を介して第１の音又は振動検出器１が設けられている。
この第１の音又は振動検出器１は、被検査体ＳＴ10の音又は振動を検出し、その音又は振動波形に対応した波形を有する第１の電気信号ＳＡ１に変換出力する機能を有している。
【０１０７】
なお、この第１の音又は振動検出器１は、以下、第１の振動検出器１として説明する。
この第１の音検出器１の出力側は、増幅器２を介して、異なる周波数帯域の周波数成分に分析する各周波数分析部３０１〜３０ｎに接続されている。
【０１０８】
第２の音又は振動検出部２０１は、図９に示すようにベースＳＴ11上に設けられ、予期できない暗騒音や暗振動、例えば台車が通過するときの音や振動、物体を落下したときの音や振動を検出し、その音又は振動波形に対応した波形を有する第２の電気信号ＳＡ２に変換出力する機能を有している。
【０１０９】
なお、この第２の音又は振動検出部２０１は、以下、第２の振動検出部２０１として説明する。
この第２の振動検出器２０１の出力側は、増幅器２０２を介して、周波数分析部３０n+1 に接続されている。
【０１１０】
一方、ＣＰＵ１４は、被検査体ＳＴ10の異常判定を行う機能を有するもので、第１の演算手段１４ａ、第２の演算手段１４ｂ及び第３の演算手段１４ｃの各機能を有している。
【０１１１】
第１の演算手段１４ａは、第１の振動検出器１及び第２の振動検出器２０１により検出される各振動のデータを各ピーク値検出回路３０１〜３０n+1 、マルチプレクサ１０及びアナログ−ディジタル変換器１１を通して取り込み、１回の検査時間内の各サンプリング期間ごとに各１個ずつ得られた最大ピーク値を大きさの順に並べ換える機能を有している。
【０１１２】
第２の演算手段１４ｂは、第１の演算手段１４ａにより大きさの順に並び換えられた各最大ピーク値について外乱内容に対応して予め定められ特定番目の最大ピーク値を有効最大ピーク値として各周波数成分ごとに抽出する機能を有している。
【０１１３】
又、この第２の演算手段１４ｂは、１回の検査時間内に各サンプリング期間ごとに各１個ずつ得られた最大ピーク値のうち、第２の振動検出器２０１により検出されたベース２０１上の外部の振動の最大ピーク値を検索し、この最大ピーク値が予め設定されたしきい値を越えた場合、この最大ピーク値を検出したサンプリング期間内にサンプリングされた被検査体ＳＴ10の振動のデータを削除するデータ削除手段としての機能を有している。
【０１１４】
この場合、第２の演算手段１４ｂは、第２の振動検出器２０１により検出された振動がしきい値より大きくなったときを含むサンプリング期間内の被検査体ＳＴ10の音を削除するとともに、その前の１又は２以上のサンプリング期間内の被検査体ＳＴ10の音を削除する。
【０１１５】
なお、第２の演算手段１４ｂは、第２の振動検出器２０１により検出された振動がしきい値より小さくなったときを含むサンプリング期間内の被検査体ＳＴ10の音を削除するとともに、その後の１又は２以上のサンプリング期間内の被検査体ＳＴ10の音を削除する。
【０１１６】
第３の演算手段１４ｃは、第２の演算手段１４ｂにより得られた有効最大ピーク値と予め設けられている設定値との比較により異常判定を行う機能を有している。
【０１１７】
次に上記の如く構成された装置の作用について説明する。
先ず、上記実施の形態の同様に、ＲＯＭ１８に後述する計算プログラムが格納されるとともに、ＲＡＭ１７に検査時間（例えば３秒）、サンプリング期間Δｔ（例えば０．１秒）、ピーク値を除外するための無効ピーク値数Ｎ_IP（例えば２個）、異常が検出されたサンプリング期間の前後の削除すべきサンプリング期間数が格納される。
【０１１８】
なお、サンプリング期間Δｔの長さは、外乱、例えばコンベア停止時の衝撃振動による影響時間より短くなるように設定するのが好ましい。
検査時間の長さは、検出すべき被検査体ＳＴ10の異常内容、例えば回転機器の回転異常によって決定する。又、異常が検出されたサンプリング期間の前後の削除すべきそれぞれのサンプリング期間数は、その外部の状態に合わせて適宜決められる。
【０１１９】
又、無効ピーク値数Ｎ_IPの数は、上記検査時間内において発生する外乱の回数を経験的に求めて、この経験的に求められた外乱の回数に基づいて決定する。
しかして、回転機器等の被検査体ＳＴ10が起動すると、第１の振動検出器１は、回転機器の振動を検出し、その振動波形に対応した波形を有する第１の電気信号ＳＡ１を出力する。
【０１２０】
この第１の電気信号ＳＡ１は、増幅器２により増幅されたのち、各バンドパスフィルタ４０１〜４０ｎに入力し、これらバンドパスフィルタ４０１〜４０ｎにより所定の異なる周波数帯域の周波数成分の信号ＳＣ１〜ＳＣｎに変換される。
【０１２１】
このうち、例えば信号ＳＣｎは、図４に示すように正弦波状波形となっている。他の信号ＳＣ１〜ＳＣ（ｎ−１）についても波形はほぼ正弦波状をなしている。
【０１２２】
これら信号ＳＣ１〜ＳＣｎは、それぞれサンプルアンドホールド回路６０１〜６０ｎ及び各比較器７０１〜７０ｎに入力される。
このうちサンプルアンドホールド回路６０ｎは、図４に示すようにピーク値を示す信号ＳＤｎを比較器７０ｎに出力する。
【０１２３】
この比較器７０ｎは、サンプルアンドホールド回路６０ｎからの信号ＳＤｎを設定値として入力するとともにバンドパスフィルタ４０ｎからの信号ＳＣｎを入力し、信号ＳＣｎより信号ＳＤｎが小さい場合には「０」を示す信号ＳＥｎをアンド回路８０ｎに出力し、かつ信号ＳＣｎより信号ＳＤｎが大きい場合には「１」を示す信号ＳＥｎをアンド回路８０ｎに出力する。
【０１２４】
図４に示す信号ＳＣｎは、例えば区間３０においては単調増加しているので 「０」を示す信号ＳＥｎがアンド回路８０ｎに出力される。
一方、アンド回路８０ｎの他方の入力端には、後述するリセット時を除いて常に「１」を示すデコーダ９からの信号ＳＦｎが出力されているので、アンド回路８０ｎからは「０」を示す信号ＳＧｎがサンプルアンドホールド回路６０ｎに出力される。
【０１２５】
このサンプルアンドホールド回路６０ｎは、アンド回路８０ｎからの信号ＳＦｎが「０」のときピーク値をサンプリングし、信号ＳＦｎ「１」のときピーク値をホールドする。
【０１２６】
従って、例えば区間３０においては、サンプルアンドホールド回路６０ｎはピーク値のサンプリング状態にあるので信号ＳＤｎも図４に示すように増加する。ところが、信号ＳＣｎの変曲点３１を境界にして信号ＳＥｎは「１」となるので、アンド回路８０ｎは「１」を示す信号ＳＧｎをサンプルアンドホールド回路６０ｎに出力し、これによって変曲点３１におけるピーク値Ｐｎ₁ がホールドされ、このピーク値Ｐｎ₁ を示す信号ＳＤｎがマルチプレクサ１０に出力される。
【０１２７】
なお、他の周波数分析部３０１〜３０（ｎ−１）からも同様にしてピーク値Ｐ₁₁…Ｐ_(n-1)1を示す各信号ＳＤ１〜ＳＤ（ｎ−１）が同時にマルチプレクサ１０に出力される。
【０１２８】
一方、第２の振動検出器２０１は、ベースＳＴ11上の外部の振動を検出し、その振動波形に対応した波形を有する第２の電気信号ＳＡ２を出力する。
この第２の電気信号ＳＡ２は、増幅器２０２により増幅されたのち、バンドパスフィルタ４０n+1 に入力し、信号ＳＣn+1 に変換される。
【０１２９】
この信号ＳＣn+1 は、サンプルアンドホールド回路６０n+1 及び比較器７０n+1 に入力される。
このうちサンプルアンドホールド回路６０n+1 は、ピーク値を示す信号ＳＤn+1 を比較器７０n+1 に出力する。
【０１３０】
この比較器７０n+1 は、サンプルアンドホールド回路６０n+1 からの信号ＳＤn+1 を設定値として入力するとともにバンドパスフィルタ４０n+1 からの信号ＳＣn+1 を入力し、信号ＳＣn+1 より信号ＳＤn+1 が小さい場合には「０」を示す信号ＳＥn+1 をアンド回路８０n+1 に出力し、かつ信号ＳＣn+1 より信号ＳＤn+1 が大きい場合には「１」を示す信号ＳＥn+1 をアンド回路８０n+1 に出力する。
【０１３１】
アンド回路８０n+1 の他方の入力端には、上記同様に、リセット時を除いて常に「１」を示すデコーダ９からの信号ＳＦn+1 が出力されているので、アンド回路８０n+1 からは「０」を示す信号ＳＧn+1 がサンプルアンドホールド回路６０n+1 に出力される。
【０１３２】
このサンプルアンドホールド回路６０n+1 は、アンド回路８０n+1 からの信号ＳＦn+1 が「０」のときピーク値をサンプリングし、信号ＳＦn+1 「１」のときピーク値をホールドする。
【０１３３】
従って、サンプルアンドホールド回路６０n+1 は、ベースＳＴ11上の外部の振動の波形に対応した信号ＳＣn+1 の変曲点におけるピーク値をホールドし、このピーク値を示す信号ＳＤn+1 をマルチプレクサ１０に出力する。
【０１３４】
一方、ＣＰＵ１４は、入出力インタフェース１６を介して所要の周波数分析部３０１〜３０ｎ、３０n+1 を指定する信号を、システムバス２０を経由してデコーダ９及びマルチプレクサ１０に出力する。
【０１３５】
この信号ＳＨがマルチプレクサ１０に送られることにより、このマルチプレクサ１０からは、図４に示すように各バンドパスフィルタ４０１〜４０ｎ、４０n+1 を通過した周波数成分の各ピーク値（例えば上記Ｐｎ₁ ）、及び外部の振動の波形に対応した信号ＳＣn+1 のピーク値を示す各信号ＳＤ１〜ＳＤｎ、ＳＤn+1 のみが、各信号ＳＩ１〜ＳＩｎ、ＳＩn+1 としてアナログ−ディジタル変換器１１に順次出力される。
【０１３６】
これら信号ＳＩ１〜ＳＩｎ、ＳＩn+1 のアナログ−ディジタル変換器１１への入力時点より時間Δｔ₁ （数１０μｓ）経過後に、ＣＰＵ１４は、図４に示すように入出力インタフェース１６を介してパルス状の信号ＳＪを時間Δｔ₂ （約３０ｍｓ以下）ごとに微分回路２２に出力する。
【０１３７】
この微分回路２２は、信号ＳＪを微分した信号ＳＫを単安定マルチバイブレータ２４及びアナログ−ディジタル変換器１１に出力する。
この信号ＳＫを入力した時点での各信号ＳＩ１〜ＳＩｎ、ＳＩn+1 は、アナログ−ディジタル変換器１１によりディジタル値の各データ信号ＳＬ１〜ＳＬｎ、ＳＬn+1 に変換され、これらデータ信号ＳＬ１〜ＳＬｎ、ＳＬn+1 は入出力インタフェース１６、ＣＰＵ１４を経てＲＡＭ１７に伝送され、このＲＡＭ１７の所定の番地にピーク値Ｐｎ₁ として格納される。
【０１３８】
なお、図４において、信号ＳＫの「１」を示す部分の前縁と信号ＳＦ１の「０」を示す部分の後縁との時間Δｔ₃ において異常検査は一時中断されるが、リアルタイムで正確にデータを収集するために、Ｓhannonのサンプリング定理によりΔｔ₃ ＜１／２Ｆ_max （ただしＦ_max は最高周波数）を満足するように設定されている。
【０１３９】
一方、信号ＳＫを入力した単安定マルチバイブレータ２４は、この信号ＳＫの後縁をトリガとして信号ＳＭをストローブ信号としてデコーダ９に出力する。
この信号ＳＭを入力したデコーダ９は、信号ＳＨが指定する各アンド回路８０１〜８０ｎ、８０n+1 に対してリセットのための「０」を示す信号ＳＦ１〜ＳＦｎ、ＳＦn+1 を出力する。
【０１４０】
図４に示すように、この「０」を示す信号ＳＦ１〜ＳＦｎ、ＳＦn+1 を入力した各アンド回路８０１〜８０ｎ、８０n+1 は「０」を示す各信号をそれぞれサンプルアンドホールド回路６０１〜６０ｎ、６０n+1 に出力し、そのピーク値は例えば図４に示す信号ＳＣの点３３まで急減し、再び前述したようにして、図４に示す次のサンプリング期間Δｔ内における最大のピーク値Ｐｎ₂ をホールドしてマルチプレクサ１０に出力する。
【０１４１】
但し、Δｔ＞Δｔ₂ ×（ｎ＋１）となるように設定されており、かつ（ｎ＋１）は周波数分析部の数である。
一方、例えば、ＲＡＭ１７にピーク値Ｐｎ₁ が格納されると、
［Δｔ−｛Δｔ₂ ×(n+1) ｝］時間後に、周波数分析部３０１を指定する信号ＳＨがデコーダ９及びマルチプレクサ１０に出力され、又、信号ＳＪが微分回路２２に出力され、周波数分析部３０ｎと同様にして、ピーク値Ｐ₁₁を示す信号ＳＩ１がアナログ−ディジタル変換器１１によりディジタル値のデータ信号ＳＬ１に変換され、このデータ信号ＳＬ１がＲＡＭ１７に伝送され、このＲＡＭ１７の所定の番地にピーク値Ｐ₁₁として格納される。
【０１４２】
この後、リセットのための「０」を示す信号ＳＦ１がアンド回路８０１に出力されサンプルアンドホールド回路６０１は再びサンプリング状態に復帰する。
このようにして、サンプリング期間Δｔ内に他の各周波数分析部３０２〜３０n+1 においてホールドされた最大のピーク値Ｐ₂₁…Ｐ_(n+1)1が順次ＲＡＭ１７の所定の番地に格納される。
【０１４３】
なお、マイクロコンピュータ１３から出力される信号ＳＪのパルス間隔は、図５に示す信号ＳＬ１〜ＳＬｎ、ＳＬn+1 に対応してＳＬ１〜ＳＬｎ、ＳＬn+1 をサンプリングする期間中はΔｔ₂ であり、例えば信号ＳＬｎから次のサンプリング期間Δｔにおける信号ＳＬn+1 をサンプリングするまでのパルス間隔は、前記したように［Δｔ−｛Δｔ₂ ×(n+1) ｝］である。
【０１４４】
このような信号ＳＪのパルス間隔を得るためのプログラムは予めＲＯＭ１８中に格納されている。さらに、Δｔ₂ ×(n+1) がサンプリング期間Δｔに等しくなるように時間Δｔ₂ を設定してもよい。
【０１４５】
上記各サンプリング期間Δｔ内において、各周波数分析部３０１〜３０ｎ、３０n+1 ごとに順次行われるＲＡＭ１７へのピーク値の格納は図５に示すようにｍ回繰り返して行われる。ＲＡＭ１７においてはピーク値は各周波数分析部３０１〜３０ｎ、３０n+1 ごとに格納する。
【０１４６】
従って、各周波数分析部３０１〜３０ｎ、３０n+1 ごとにｍ個のデータが格納される。
しかして、ＲＡＭ１７に格納されたデータ数がｍ×(n+1) （ただし、ｍはサンプリング回数、(n+1) は周波数分析部の数である）に達すると、ＣＰＵ１４の第１の演算手段１４ａは、各周波数分析部３０１〜３０ｎ、３０n+1 ごとにピーク値を大きい順に並び換える演算を行い、この演算結果を図６に示すようにＲＡＭ１７に格納する。
【０１４７】
次にＣＰＵ１４の第２の演算手段１４ｂは、例えばコンベア停止時の衝撃振動、エアシリンダ動作時の衝撃振動などの被検査体ＳＴ10以外の振動源からの外乱の影響を除外するために、各周波数成分ごとにＲＡＭ１７に格納されている上記大きい順に並べられたピーク値の大きい側からＮ_IP個、例えば２個のデータを除外し、Ｎ_IP＋１番目のデータを有効最大ピーク値Ｐ_EM1 …Ｐ_EMn として抽出しＲＡＭ１７に格納する。
【０１４８】
又、この第２の演算手段１４ｂは、１回の検査時間内に各サンプリング期間ごとに各１個ずつ得られた最大ピーク値のうち、第２の音検出器２０１により検出されたベースＳＴ11上の外部の振動の最大ピーク値を検索する。
【０１４９】
そして、第２の演算手段１４ｂは、この最大ピーク値が予め設定されたしきい値を越えたか否かを判断し、最大ピーク値がしきい値を越えた場合、この最大ピーク値を検出したサンプリング期間内にサンプリングされた被検査体ＳＴ10の音の振幅のデータを削除する。
【０１５０】
この場合、第２の演算手段１４ｂは、第２の振動検出器２０１により検出された振動がしきい値より大きくなったときを含むサンプリング期間内の被検査体ＳＴ10の音を削除するとともに、その前の１又は２以上のサンプリング期間内の被検査体ＳＴ10の音を削除する。
【０１５１】
例えば、被検査体ＳＴ10を載置するベースＳＴ11の近傍に、例えば台車が通過したり、又はベースＳＴ11の近傍で物体が落下すると、第２の振動検出器２０１は、台車の通過するときの振動又は物体の落下したときの振動を検出し、これら振動の波形に対応した電気信号ＳＡ２を出力する。
【０１５２】
一方、第１の振動検出器１から出力される電気信号ＳＡ１においても台車の通過の振動、又は物体の落下の振動による波形Ｇｂが現れる。
従って、第２の演算手段１４ｂは、サンプリング期間Ｔａ内において第２の振動検出器２０１により検出されたベースＳＴ11上の振動の最大ピーク値を検索し、この最大ピーク値が予め設定されたしきい値を越えた場合、この最大ピーク値を検出したサンプリング期間内にサンプリングされた被検査体ＳＴ10の振動のデータを削除する。
【０１５３】
次にＣＰＵ１４の第３の演算手段１４ｃは、各周波数成分ごとに、予めＲＡＭ１７に設定されている各設定値Ｐ_T1…Ｐ_Tnと、これら各設定値Ｐ_T1…Ｐ_Tnに対応する各有効最大ピーク値Ｐ_EM1 …Ｐ_EMn とを比較し、有効最大ピーク値Ｐ_EM1 …Ｐ_EMn のうち一つでも設定値を越えていれば「被検査体ＳＴ10に異常あり」と判定する。
【０１５４】
このように上記第２の実施の形態においては、被検査体ＳＴ10に対して外部の音を第２の振動検出器２０１により検出し、この検出された振動が予め設定されたしきい値を越えると、少なくともサンプリング中の被検査体ＳＴ10の振動のデータを削除するので、被検査体ＳＴ10の振動の影響を受けることなく、予期せぬ衝撃性の暗騒音や暗振動を確実に検出して、確実に被検査体ＳＴ10の異常判定ができる。
(3) 次に本発明の第３の実施の形態について説明する。なお、上記第１の実施の形態と同一部分についてはその説明を省略する。
【０１５５】
この発明の異常検査方法は、被検査体の音又は機械的振動を検出し、この音又は機械的振動を各周波数成分に分けてこれら成分における最大ピーク値を抽出し、これら最大ピーク値を大きさの順に並び換え、その中の特定番目の最大ピーク値に基づいて被検査体の異常判定を行う異常検査方法において、
特定番目の最大ピーク値として、各周波数成分における全サンプル数に対し所定の割合で選択されたサンプルの中の最大ピーク値を用いるものである。
【０１５６】
ここで、所定の割合は、各周波数成分毎に設定される。
具体的に説明すると、上記第１の実施の形態では、第１の演算手段１４ａにより各周波数分析部３０１〜３０ｎ、３０n+1 ごとにピーク値を大きい順に並び換える演算を行い、この演算結果を図６に示すようにＲＡＭ１７に格納する。
【０１５７】
続いて、第２の演算手段１４ｂは、例えばコンベア停止時の衝撃振動、エアシリンダ動作時の衝撃振動などの被検査体以外の振動源からの外乱の影響を除外するために、各周波数成分ごとにＲＡＭ１７に格納されている上記大きい順に並べられたピーク値の大きい側からＮ_IP個、例えば２個のデータを除外し、Ｎ_IP＋１番目のデータを有効最大ピーク値Ｐ_EM1 …Ｐ_EMn として抽出しＲＡＭ１７に格納する。
【０１５８】
この第２の演算手段１４ｂは、１回の検査時間内に各サンプリング期間ごとに各１個ずつ得られた最大ピーク値のうち、第２の音検出器２０１により検出された防音室ＳＴ２の外部の音の振幅の最大ピーク値を検索する。
【０１５９】
そして、この第２の演算手段１４ｂは、この最大ピーク値が予め設定されたしきい値を越えたか否かを判断し、最大ピーク値がしきい値を越えた場合、この最大ピーク値を検出したサンプリング期間内にサンプリングされた被検査体ＳＴ１の音の振幅のデータを削除する。
【０１６０】
この場合、第２の演算手段１４ｂは、第２の音検出器２０１により検出された音がしきい値より大きくなったときを含むサンプリング期間内の被検査体ＳＴ１の音を削除するとともに、その前の１又は２以上のサンプリング期間内の被検査体ＳＴ１の音を削除する。
【０１６１】
又、第２の演算手段１４ｂは、第２の音検出器２０１により検出された音がしきい値より小さくなったときを含むサンプリング期間内の被検査体ＳＴ１の音を削除するとともに、その後の１又は２以上のサンプリング期間内の被検査体ＳＴ１の音を削除する。
【０１６２】
これに対して第３の実施の形態では、回転機器等の被検査体の音又は機械的振動を検出し、この音又は機械的振動を各周波数成分に別けてこれら成分における最大ピーク値を抽出し、これら最大ピーク値を大きさの順に並び換えて上記図６と同様なデータ列とする。
【０１６３】
十分防音対策された防音室内や十分振動絶縁された被検査体上であれば、希に発生する衝撃波或いは突発性の暗騒音や暗振動は、被検査体の音又は振動よりも大きい。
【０１６４】
そこで、データ列の全サンプル数に対し所定の割合で大きい方からデータを削除し、残ったデータの中の最大ピーク値に基づいて被検査体の異常判定を行う。すなわち、最も大きい方から次式で得られた除外データ数を削除して、衝撃性或いは突発性の暗騒音や暗振動の影響を受けない有効最大ピーク値を得る。ここで、指定値は、除外データ数を全サンプル数の割合（百分率）とする。
【０１６５】
除外データ数＝全サンプル数−（全サンプル数×指定値）／１００ …(4)
なお、除外するデータを決めるのに用いる所定の割合は、例えば書き換え可能な記録媒体に予め記録しておき、検査の際に、記録媒体からその割合を呼び出して実際に除外されるデータ数を計算し、この値を用いて測定データ処理を行い除外した後の残りのデータを用いて異常判定を行う。
【０１６６】
この所定の割合は、各周波数成分毎に設定することができる。この所定の割合としては、例えば除外する分を全サンプル数の３０％程度以下にする。
そして、上記指定値は、書き換え可能な記録媒体に格納しておき、本プログラムの実行開始直後に前記計算を１回行い、各周波数毎の除外データ数をＲＡＭ上に記録する。
【０１６７】
以降、検査する毎にＲＡＭ上に記録された除外データ数を使って有効最大ピーク値を得て、これと予め設定された判定基準とを比較して異常判定を行う。
このように上記第３の実施の形態によれば、回転機器等の被検査体の音又は機械的振動を検出し、この音又は機械的振動を各周波数成分に別けてこれら成分における最大ピーク値を抽出し、これら最大ピーク値を大きさの順に並び換えてデータ列とし、このデータ列の全サンプル数に対し所定の割合で大きい方からデータを削除し、残ったデータの中の最大ピーク値に基づいて被検査体の異常判定を行うようにしたので、上記第１の実施の形態と同様の効果を奏することは言うまでもなく、そのうえ異常音又は異常振動の減衰時間は各周波数成分毎に異なるので、これまでのように特定する順位（特定番目）を減衰時間が長い低周波数域で合わせると減衰時間が短い高周波数域では必要異常にデータを削減してしまう場合があり、一方それを減衰時間が短い高周波数域で合わせると暗騒音や暗振動がその低周波数域では十分に減衰しないので不要なデータが残る場合がある。これらはともに誤判定を生じてしまう原因となるが、上記第３の実施の形態によれば、そうした誤判定を防止できるので、より信頼性の高い異常検出が可能となる。
【０１６８】
なお、本発明は、上記第１乃至第３の実施の形態に限定されるものでなく次の通り変形してもよい。
例えば、異常検査装置は被検査体として回転機器に限ることなく、例えば原子炉、製造設備等の異常検出にも用いることができる。このとき、外乱の影響時間の長短に応じて、サンプリング期間Δｔ及び無効ピーク値数Ｎ_IPを適宜変更する必要がある。
【０１６９】
さらに、表示部としてＣＲＴを用いているが、印字装置のようなものでもよい。
又、各ピーク検出回路５０１〜５０n+1 は、実効値検出回路であってもよく、或いは正半波ピーク値をホールドしただけでなく負半波ピーク値もホールドして絶対値ピーク値とする検出回路であってもよい。又、第２の音又は振動検出器２０１の出力信号を利用して、衝撃性又は突発性の暗騒音や暗振動が入ってきたときはホールド状態になるように各ピーク検出回路５０１〜５０n+1 の各サンプルアンドホールド制御信号を直接制御してもよい。
【０１７０】
図１０はかかる衝撃性又は突発性の暗騒音や暗振動が入ってきたときにホールド状態とする場合の構成図である。第２の音又は振動検出器２０１の出力側は、増幅器２０２、バンドパスフィルタ２０３、ピーク値又は実効値検出回路２０４、比較回路２０５を介して、各周波数分析部５０n+1 に接続されている。
【０１７１】
この周波数分析部５０n+1 は、サンプルアンドホールド回路６０n+1 と、比較器７０n+1 と、アンド回路８０n+1 と、これら比較器７０n+1 とアンド回路８０n+1 との間に接続されたアンド回路９０n+1 とから構成されている。
【０１７２】
このうちアンド回路９０n+1 の入力部には、第２の音又は振動検出器２０１からの電気信号が入力している。
又、上記第１〜第３の実施の形態では、被検査体の音又は機械的振動を検出し、この音又は機械的振動を各周波数成分に分けてこれら成分における最大ピーク値を抽出し、これら最大ピーク値を大きさの順に並び換え、予め定められた特定番目の最大ピーク値に基づいて被検査体の異常判定を行う場合について説明したが、これは被検査体の音又は機械的振動を検出し、この音又は機械的振動を各周波数成分に分けてこれら成分における最大ピーク値を抽出して被検査体の異常判定を行う場合でも、被検査体に対して外部の音又は機械的振動を検出し、この検出された外部の音又は機械的振動が予め設定されたしきい値を越えた場合、少なくともサンプリング中の被検査体の音又は機械的振動を削除することにより、信頼性の高い異常検出ができる。
【０１７３】
【発明の効果】
以上詳記したように本発明によれば、最大ピーク値を検出したサンプリング期間内で第１の音又は振動検出器により検出された被検査体の音又は機械的振動を削除し、かつ少なくとも最大ピーク値を検出したサンプリング期間後の１又は２以上のサンプリング期間内で第１の音又は振動検出器により検出された被検査体の音又は機械的振動を削除するので、予期できない衝撃性或いは突発性の暗騒音や暗振動による誤判定を防止して確実に被検査体の異常判定ができる信頼性の高い異常検査方法及びその装置を提供できる。
【図面の簡単な説明】
【図１】本発明に係わる異常検査装置の第１の実施の形態を示す構成図。
【図２】第１及び第２の音又は振動検出器の配置図。
【図３】異常検出動作のフローチャート。
【図４】異常検出動作のタイミング図。
【図５】本装置の演算制御部に逐次入力する最大ピーク値を示す図。
【図６】本装置の演算制御部のＲＡＭにて大きい順に並べ換えられた最大ピーク値の模式図。
【図７】被検査体の音及びその外部音の信号波形を示す図。
【図８】検査結果の表示例を示す図。
【図９】本発明に係わる異常検査装置の第２の実施の形態を示す構成図。
【図１０】本発明装置の変形例を示す構成図。
【符号の説明】
１…第１の音又は振動検出器、
ＳＴ１…被検査体、
３０１〜３０ｎ，３０n+1 …周波数分析部、
２０１…第２の音又は振動検出部、
５０１〜５０ｎ，５０n+1 …ピーク値検出回路、
１０…マルチプレクサ、
１３…マイクロコンピュータ、
１４ａ…第１の演算手段、
１４ｂ…第２の演算手段、
１４ｃ…第３の演算手段。[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an abnormality inspection method and apparatus for detecting an abnormality of various devices based on sound or mechanical vibration.
[0002]
[Prior art]
For example, in a manufacturing process of a rotating device, a method of automatically performing mechanical comprehensive evaluation by vibration analysis has been developed. A typical example will be described next.
The first method is to detect mechanical vibration or sound generated in an object to be inspected, such as a rotating device, and to analyze the power spectrum or 1/3 octave frequency of the detected signals to obtain a pattern of normal and abnormal products. Are compared.
[0003]
The second method uses a high-pass filter to extract a high-frequency component of about 1 kHz or more from a mechanical vibration or sound detection signal generated in the object to be inspected. Are compared.
[0004]
The third method is a method in which an envelope signal is extracted from the high frequency signal obtained in the same manner as the second method by an envelope detection circuit, and the normal product and the abnormal product are compared and determined.
[0005]
However, in the first method, since the average power during measurement is obtained, it is shocking due to the ball cage of the bearing portion of the object to be measured, the rotation abnormality due to the presence of dust, the resonance abnormality of each component, etc. It is difficult to detect a pulsed signal that occurs only occasionally.
[0006]
In contrast, in the second and third methods, it is possible to detect an abnormal signal that cannot be detected by the first method, but on the other hand, various types of vibrations that occur in vibration sources other than the object to be inspected during actual inspection. It has a defect that a normal product is erroneously determined as an abnormal product due to the influence of disturbance such as sound and mechanical vibration. In addition, it is difficult to detect subtle bearing defects.
[0007]
That is, noise and vibration generated in the production line are mixed in the sound or vibration detection signal obtained by detecting the object to be inspected, and it is impossible to detect subtle abnormal sounds.
Therefore, Japanese Patent Publication No. 4-4534 discloses a technique in which the background noise and dark vibration are removed to enhance the detection capability of abnormal noise.
[0008]
This technology detects the sound or vibration of the object to be inspected, extracts each frequency component from this sound or vibration detection signal through, for example, each bandpass filter for each frequency component, and samples it within an inspection time consisting of a plurality of sampling periods. The maximum peak value for each frequency component in the period is detected one by one, and this is digitally converted and stored in the arithmetic control unit.
[0009]
Then, disturbance data is removed by rearranging the stored maximum peak values in order of magnitude for each frequency component, and the rearranged maximum peak value is determined to be a predetermined specific maximum corresponding to the content of the disturbance. The peak value is extracted for each frequency component as an effective maximum peak value for abnormality detection, and the abnormality of the inspected object is determined by comparing the effective maximum peak value for each frequency component with a preset value. To do.
[0010]
[Problems to be solved by the invention]
However, in the technique disclosed in Japanese Examined Patent Publication No. 4-4534, the arrangement of the maximum peak values for each frequency component obtained by removing disturbance data over the inspection time is rearranged in order of magnitude, and the order is determined in descending order. Since the specific data is used as an effective maximum peak value for anomaly determination, it is possible to eliminate regularly occurring impulsive background noise and background vibrations, but unforeseen background noise and background vibrations such as carts pass through. It is difficult to set the specific number of data to be removed for the sound, vibration, and sound and vibration when the object is dropped.
[0011]
For this reason, the specific number of data for removing background noise and background vibration is a number that allows for a margin, and as a result, the ability to determine abnormality of the object to be inspected is reduced.
[0012]
Accordingly, an object of the present invention is to provide a highly reliable abnormality inspection method capable of reliably determining an abnormality of an object to be inspected by preventing an erroneous determination due to unexpected impact or sudden background noise or dark vibration. .
[0013]
Another object of the present invention is to provide a highly reliable abnormality inspection apparatus that can prevent an erroneous determination due to unexpected impact or sudden background noise or dark vibration and reliably determine the abnormality of an object to be inspected. To do.
[0014]
[Means for Solving the Problems]
The present invention , Housed in a soundproof room The sound or mechanical vibration of the inspected object By the first sound or vibration detector arranged in the soundproof room Detecting, dividing sound or mechanical vibration into each frequency component, extracting each maximum peak value in these frequency components, rearranging these maximum peak values in order of magnitude, to a predetermined specific maximum peak value In an abnormality inspection method for determining abnormality of an object to be inspected, external sound or mechanical vibration is applied to the object to be inspected. By a second sound or vibration detector placed outside the soundproof room Detect this detected external sound or mechanical vibration Maximum peak value of When exceeds the preset threshold, Detected by the first sound or vibration detector within the sampling period in which the maximum peak value was detected Delete the sound or mechanical vibration of the object under test And the sound or mechanical vibration of the object detected by the first sound or vibration detector within one or more sampling periods after the sampling period in which at least the maximum peak value is detected is deleted. This is an abnormality inspection method.
[0015]
The present invention , The sound or mechanical vibration of the object to be inspected housed in the soundproof room is detected by the first sound or vibration detector arranged in the soundproof room, and the sound or mechanical vibration is divided into the respective frequency components. Extract each maximum peak value, rearrange these maximum peak values in order of magnitude, and place them outside the soundproof room in an abnormality inspection device that performs abnormality determination of the inspected object based on a predetermined specific maximum peak value A second sound or vibration detector that detects external sound or mechanical vibration with respect to the object to be inspected, and a maximum peak of external sound or mechanical vibration detected by the second sound or vibration detector When the value exceeds a preset threshold, the sound or mechanical vibration of the object to be inspected detected by the first sound or vibration detector is deleted within the sampling period in which the maximum peak value is detected. And at least And and a data deletion means for deleting a sound or mechanical vibration of the object to be inspected which has been detected by the first sound or vibration detectors in one or more of the sampling period after the sampling period to detect the maximum peak value It is an abnormality inspection device.
[0027]
DETAILED DESCRIPTION OF THE INVENTION
(1) A first embodiment of the present invention will be described below with reference to the drawings.
The abnormality inspection method of the present invention detects the sound or mechanical vibration of an object to be inspected such as a rotating device, divides this sound or mechanical vibration into each frequency component, and extracts the maximum peak value in these components to be inspected. In an abnormality inspection method for performing abnormality determination of a body, more specifically, performing abnormality determination of an object to be inspected based on a specific maximum peak value obtained by rearranging these maximum peak values in order of size,
If an external sound or mechanical vibration is detected for the object to be inspected, and the detected external sound or mechanical vibration exceeds a preset threshold value, at least the sound of the object to be inspected during sampling is detected. Alternatively, mechanical vibration is eliminated.
[0028]
FIG. 1 is a configuration diagram of an abnormality inspection apparatus to which such an abnormality inspection method is applied.
As shown in FIG. 2, the first sound or vibration detector 1 detects the sound or vibration of the inspected object ST1 such as a rotating device, and the first electric signal SA1 having a waveform corresponding to the sound or vibration waveform. It has the function of converting to output.
[0029]
The first sound or vibration detector 1 will be described below as the first sound detector 1.
The inspected object ST1 is accommodated in the soundproof room ST2 via each vibration isolating rubber ST3.
[0030]
The first sound detector 1 is disposed in the soundproof room ST2, and detects the amplitude of the sound of the subject ST1 in the soundproof room ST2.
The output side of the first sound detector 1 is connected via the amplifier 2 to each frequency analysis unit 301 to 30n that analyzes frequency components in different frequency bands. However, in this embodiment, n = 28 (hereinafter the same).
[0031]
Each of these frequency analysis units 301 to 30n is connected to the amplifier 2 directly, for example, each band pass for obtaining frequency components of 28 frequency bands for each 1/3 octave whose center frequency is 20 Hz to 10000 Hz. The filters 401 to 40n and the peak value detection circuits 501 to 50n connected to the output sides of the bandpass filters 401 to 40n, respectively.
[0032]
Among these, each of the peak value detection circuits 501 to 50n includes sample and hold circuits 601 to 60n, comparators 701 to 70n, and AND circuits 801 to 80n, respectively.
[0033]
These sample and hold circuits 601 to 60n each have a function of holding the positive half-wave peak value, and are connected to the output sides of the bandpass filters 401 to 40n, respectively.
[0034]
Each of the comparators 701 to 70n has an output side of each of the bandpass filters 401 to 40n connected to each input side, and an output side of each of the sample and hold circuits 601 to 60n connected to each input side.
[0035]
Each AND circuit 801 to 80n has an output side connected to the control input side of each sample and hold circuit 601 to 60n, and an input side connected to the output side of each comparator 701 to 70n.
[0036]
The other input side of these AND circuits 801 to 80n is connected to the output side of the decoder 9 described later.
Each frequency analysis unit 301 to 30 n having such a configuration is connected to the input side of the multiplexer 10.
[0037]
On the other hand, the second sound or vibration detection unit 201 is arranged outside the soundproof room ST2 as shown in FIG. 2, and unexpected background noise and dark vibration, for example, sound and vibration when a carriage passes, drop an object It has a function of detecting the sound and vibration at the time, and converting and outputting the second electric signal SA2 having a waveform corresponding to the sound or vibration waveform.
[0038]
The second sound or vibration detection unit 201 will be described below as the second sound detection unit 201.
The output side of the second sound detector 201 is connected to the frequency analysis unit 30n + 1 via the amplifier 202.
[0039]
The frequency analysis unit 30n + 1 includes a bandpass filter 40n + 1 that is directly electrically connected to the amplifier 202, and a peak value detection circuit 50n + 1 that is separately connected to the output side of the bandpass filter 40n + 1. It consists of and.
[0040]
Of these, the bandpass filter 40n + 1 is set so that its frequency band covers the entire frequency band of each of the bandpass filters 401 to 40n used for abnormality determination.
[0041]
The peak value detection circuit 50n + 1 includes a sample and hold circuit 60n + 1, a comparator 70n + 1 and an AND circuit 80n + 1.
The sample and hold circuit 60n + 1 has a function of holding the positive half-wave peak value, and is connected to the output side of the bandpass filter 40n + 1.
[0042]
In the comparator 70n + 1, the output side of the bandpass filter 40n + 1 is connected to each input side, and the output side of the sample and hold circuit 60n + 1 is connected to each input side.
[0043]
The output side of the AND circuit 80n + 1 is connected to the control input side of the sample and hold circuit 60n + 1, and the input side is connected to the output side of the comparator 70n + 1.
[0044]
The other input side of the AND circuit 80n + 1 is connected to the output side of the decoder 9 described later.
The frequency analyzer 30n + 1 having such a configuration is connected to the input side of the multiplexer 10.
[0045]
The output side of the multiplexer 10 is connected from the analog-digital converter 11 to the microcomputer 13 via the system bus 12.
The microcomputer 13 includes a CPU (Central Processing Unit) 14, an input / output interface 16 connected to the CPU 14 via a system bus 15, and a RAM (Random Access Memory) 17 capable of writing and reading data. , A ROM (Read Only Memory) 18 in which a control program is stored, and a CRT (Cathode Ray Tube) interface 19.
[0046]
Of these, the CPU 14 has a function of determining abnormality of the object to be inspected, and has the functions of the first computing means 14a, the second computing means 14b, and the third computing means 14c.
[0047]
The first computing means 14a converts the amplitude data of the sounds detected by the first sound detector 1 and the second sound detector 201 into the peak value detection circuits 301 to 30n + 1, the multiplexer 10 and the analog-digital. It has a function of taking in through the converter 11 and rearranging the maximum peak values obtained one by one for each sampling period within one inspection time in order of size.
[0048]
The second calculating means 14b sets each of the maximum peak values rearranged in order of magnitude by the first calculating means 14a in advance corresponding to the content of the disturbance, and sets each specific peak value as an effective maximum peak value. It has a function of extracting each frequency component.
[0049]
In addition, the second calculation means 14b has the soundproof room ST2 detected by the second sound detector 201 among the maximum peak values obtained one by one for each sampling period within one inspection time. When the maximum peak value of the amplitude of the external sound is retrieved and the maximum peak value exceeds a preset threshold value, the object ST1 sampled within the sampling period in which the maximum peak value is detected It has a function as a data deleting means for deleting sound amplitude data.
[0050]
In this case, the second calculation means 14b deletes the sound of the inspected object ST1 within the sampling period including the time when the sound detected by the second sound detector 201 becomes larger than the threshold value. It has a function as data deletion means for deleting the sound of the inspected object ST1 within the previous one or more sampling periods.
[0051]
The second computing means 14b deletes the sound of the object ST1 within the sampling period including the time when the sound detected by the second sound detector 201 becomes smaller than the threshold value, and thereafter It has a function of deleting the sound of the subject ST1 within one or more sampling periods.
[0052]
The third calculation means 14c has a function of performing abnormality determination by comparing the effective maximum peak value obtained by the second calculation means 14b with a preset value.
[0053]
A part of the RAM 17 is composed of a nonvolatile RAM for storing set values and normal patterns.
Although not shown, the microcomputer 13 is provided with an input unit for allowing the RAM 17 and ROM 18 to store the control program, setting values, and the like from the outside.
[0054]
The input / output interface 16 is connected to the analog-digital converter 11 via the system bus 12. Further, the input / output interface 16 is connected to the decoder 9 and the multiplexer 10 via the system bus 20.
[0055]
The input / output interface 16 is connected to the input side of the differentiation circuit 22 via a line 21.
The output side of the differentiating circuit 22 is connected to the input side of the analog-digital converter 11 and the monostable multivibrator 24 via a strobe line 23.
[0056]
The output side of the monostable multivibrator 24 is connected to the decoder 9 via a strobe line 25.
These decoder 9, differentiation circuit 22 and monostable multivibrator 24 constitute a timing circuit 26.
[0057]
Further, the timing circuit 26 and the microcomputer 13 constitute an arithmetic control unit 27.
The CRT interface 19 is connected to a CRT display 28 for displaying inspection results.
[0058]
Next, the operation of the apparatus configured as described above will be described according to the abnormality inspection flowchart shown in FIG.
First, in step # 1, a calculation program (to be described later) is stored in the ROM 18, and an inspection time (for example, 3 seconds), a sampling period Δt (for example, 0.1 seconds), and an invalid peak value for excluding peak values are stored in the RAM 17. Number N _IP Stores the number of sampling periods to be deleted before and after the sampling period in which an abnormality is detected (for example, two).
[0059]
Note that the length of the sampling period Δt is preferably set to be shorter than the influence time due to disturbance, for example, impact vibration when the conveyor is stopped.
The length of the inspection time is determined by the abnormal content of the inspection object to be detected, for example, the rotational abnormality of the rotating device.
[0060]
The number of invalid peak values N _IP The number of disturbances is determined based on the number of disturbances empirically obtained within the inspection time and the number of disturbances obtained empirically.
Thus, the first sound detector 1 is brought into contact with the rotating device that is the device under test 1, and the rotating device is activated.
[0061]
The first sound detector 1 detects the sound of the rotating device and outputs a first electric signal SA1 having a waveform corresponding to the amplitude waveform of this sound.
The first electric signal SA1 is amplified by the amplifier 2 and then input to the band-pass filters 401 to 40n. The band-pass filters 401 to 40n convert the first electric signals SA1 into signals SC1 to SCn of frequency components in predetermined different frequency bands. Converted.
[0062]
Hereinafter, the signal processing method of the signal SCn mainly in the case of n = 28, that is, 28 channels (center frequency is 10000 Hz) will be described in detail based on FIG. 1 and FIG.
[0063]
The signal SCn has a sinusoidal waveform as shown in FIG. The waveforms of the other signals SC1 to SC (n-1) are almost sinusoidal.
These signals SC1 to SCn are input to sample and hold circuits 601 to 60n and comparators 701 to 70n, respectively.
[0064]
Among them, the sample and hold circuit 60n outputs a signal SDn indicating a peak value to the comparator 70n as shown in FIG.
The comparator 70n receives the signal SDn from the sample and hold circuit 60n as a set value and also receives the signal SCn from the bandpass filter 40n. When the signal SDn is smaller than the signal SCn, the comparator 70n indicates “0”. SEn is output to the AND circuit 80n, and when the signal SDn is larger than the signal SCn, the signal SEn indicating "1" is output to the AND circuit 80n.
[0065]
Since the signal SCn shown in FIG. 4 monotonously increases in the interval 30, for example, the signal SEn indicating “0” is output to the AND circuit 80n.
On the other hand, since the signal SFn from the decoder 9 that always indicates “1” is output to the other input terminal of the AND circuit 80n except at the time of reset described later, the signal that indicates “0” is output from the AND circuit 80n. SGn is output to the sample and hold circuit 60n.
[0066]
The sample and hold circuit 60n samples the peak value when the signal SFn from the AND circuit 80n is “0”, and holds the peak value when the signal SFn is “1”.
[0067]
Therefore, for example, in the section 30, since the sample and hold circuit 60n is in the peak value sampling state, the signal SDn also increases as shown in FIG.
However, since the signal SEn becomes “1” at the inflection point 31 of the signal SCn, the AND circuit 80n outputs the signal SGn indicating “1” to the sample-and-hold circuit 60n, thereby the inflection point 31. Peak value Pn at ₁ Is held, and this peak value Pn ₁ Is output to the multiplexer 10.
[0068]
Although not shown in FIG. 3, the peak value P is similarly obtained from the other frequency analysis units 301 to 30 (n−1). ₁₁ ... P _{(n-1) 1} Are output to the multiplexer 10 simultaneously.
[0069]
On the other hand, the second sound detector 201 detects a sound outside the soundproof room ST2, and outputs a second electric signal SA2 having a waveform corresponding to the amplitude waveform of the sound.
The second electric signal SA2 is amplified by the amplifier 202, and then input to the band pass filter 40n + 1 to be converted into a signal SCn + 1.
[0070]
This signal SCn + 1 is input to the sample and hold circuit 60n + 1 and the comparator 70n + 1.
Among these, the sample and hold circuit 60n + 1 outputs a signal SDn + 1 indicating a peak value to the comparator 70n + 1.
[0071]
The comparator 70n + 1 inputs the signal SDn + 1 from the sample and hold circuit 60n + 1 as a set value and also receives the signal SCn + 1 from the bandpass filter 40n + 1, and receives a signal from the signal SCn + 1. When SDn + 1 is small, the signal SEn + 1 indicating "0" is output to the AND circuit 80n + 1, and when the signal SDn + 1 is larger than the signal SCn + 1, the signal SEn + indicating "1" is output. 1 is output to the AND circuit 80n + 1.
[0072]
Since the signal SFn + 1 from the decoder 9 always indicating “1” is output to the other input terminal of the AND circuit 80n + 1 except for the time of reset, the AND circuit 80n + 1 receives from the AND circuit 80n + 1. A signal SGn + 1 indicating "0" is output to the sample and hold circuit 60n + 1.
[0073]
The sample and hold circuit 60n + 1 samples the peak value when the signal SFn + 1 from the AND circuit 80n + 1 is "0", and holds the peak value when the signal SFn + 1 is "1".
[0074]
Accordingly, the sample-and-hold circuit 60n + 1 holds the peak value at the inflection point of the signal SCn + 1 corresponding to the waveform of the sound amplitude outside the soundproof room ST2, and the signal SDn + 1 indicating this peak value is held. Output to the multiplexer 10.
[0075]
On the other hand, the CPU 14 outputs signals specifying the required frequency analysis units 301 to 30n and 30n + 1 via the input / output interface 16 to the decoder 9 and the multiplexer 10 via the system bus 20.
[0076]
When this signal SH is sent to the multiplexer 10, the multiplexer 10 causes each peak value (for example, the above Pn) of the frequency components that have passed through the bandpass filters 401 to 40n and 40n + 1 as shown in FIG. ₁ ) And only the signals SD1 to SDn and SDn + 1 indicating the peak value of the signal SCn + 1 corresponding to the waveform of the amplitude of the external sound are converted into the signals SI1 to SIn and SIn + 1 as the analog-digital converter 11. Are output sequentially.
[0077]
Time Δt from the time when these signals SI1 to SIn and SIn + 1 are input to the analog-to-digital converter 11 ₁ After the lapse of (several tens of μs), the CPU 14 applies the pulse signal SJ to the time Δt via the input / output interface 16 as shown in FIG. ₂ It outputs to the differentiation circuit 22 every (about 30 ms or less).
[0078]
The differentiating circuit 22 outputs a signal SK obtained by differentiating the signal SJ to the monostable multivibrator 24 and the analog-digital converter 11.
The signals SI1 to SIn and SIn + 1 at the time when the signal SK is input are converted into digital data signals SL1 to SLn and SLn + 1 by the analog-digital converter 11, and these data signals SL1 to SLn are converted. , SLn + 1 is transmitted to the RAM 17 via the input / output interface 16 and the CPU 14, and the peak value Pn is set at a predetermined address of the RAM 17 in step # 2. ₁ Stored as
[0079]
In FIG. 4, the time Δt between the leading edge of the portion indicating “1” of the signal SK and the trailing edge of the portion indicating “0” of the signal SF1. _Three In order to collect data accurately in real time, Shannon's sampling theorem indicates that Δt _Three <1 / 2F _max (However F _max Is set to satisfy the highest frequency).
[0080]
On the other hand, the monostable multivibrator 24 having received the signal SK outputs the signal SM to the decoder 9 as a strobe signal using the trailing edge of the signal SK as a trigger.
The decoder 9 receiving the signal SM outputs signals SF1 to SFn and SFn + 1 indicating "0" for resetting to the AND circuits 801 to 80n and 80n + 1 designated by the signal SH.
[0081]
As shown in FIG. 4, the AND circuits 801 to 80n and 80n + 1, to which the signals SF1 to SFn and SFn + 1 indicating "0" are input, respectively, indicate the signals indicating "0" to the sample and hold circuits 601 to 601, respectively. 60n, 60n + 1, and the peak value rapidly decreases to, for example, the point 33 of the signal SC shown in FIG. 4, and again as described above, the maximum peak value Pn within the next sampling period Δt shown in FIG. ₂ Is held and output to the multiplexer 10.
[0082]
However, Δt> Δt ₂ X (n + 1) is set, and (n + 1) is the number of frequency analysis units.
On the other hand, for example, the peak value Pn is stored in the RAM 17. ₁ Is stored,
[Δt− {Δt ₂ × (n + 1)}] After a time, the signal SH designating the frequency analysis unit 301 is output to the decoder 9 and the multiplexer 10, and the signal SJ is output to the differentiation circuit 22 in the same manner as the frequency analysis unit 30n. , Peak value P ₁₁ Is converted into a digital data signal SL1 by the analog-to-digital converter 11, and this data signal SL1 is transmitted to the RAM 17, where the peak value P is stored at a predetermined address in the RAM 17. ₁₁ Stored as
[0083]
Thereafter, a signal SF1 indicating “0” for resetting is output to the AND circuit 801, and the sample and hold circuit 601 returns to the sampling state again.
In this way, the maximum peak value P held in each of the other frequency analysis units 302 to 30n + 1 within the sampling period Δt. _{twenty one} ... P _{(n + 1) 1} Are sequentially stored in a predetermined address of the RAM 17.
[0084]
Note that the pulse interval of the signal SJ output from the microcomputer 13 is Δt during the period of sampling SL1 to SLn and SLn + 1 corresponding to the signals SL1 to SLn and SLn + 1 shown in FIG. ₂ For example, the pulse interval from the signal SLn to the sampling of the signal SLn + 1 in the next sampling period Δt is [Δt− {Δt ₂ × (n + 1)}].
[0085]
A program for obtaining such a pulse interval of the signal SJ is stored in the ROM 18 in advance. Furthermore, Δt ₂ Time Δt so that x (n + 1) is equal to the sampling period Δt. ₂ May be set.
[0086]
During each sampling period Δt, the peak value is stored in the RAM 17 sequentially for each frequency analysis unit 301 to 30n, 30n + 1, and is repeated m times as shown in FIG. In the RAM 17, the peak value is stored for each frequency analysis unit 301 to 30n, 30n + 1.
[0087]
Accordingly, m pieces of data are stored for each of the frequency analysis units 301 to 30n and 30n + 1.
In step # 3, when the number of data stored in the RAM 17 reaches m × (n + 1) (where m is the number of samplings and (n + 1) is the number of frequency analysis units), the CPU 14 In step # 4, the first calculation means 14a performs a calculation for rearranging the peak values in descending order for each of the frequency analysis units 301 to 30n and 30n + 1, and the calculation result is stored in the RAM 17 as shown in FIG. Store.
[0088]
Next, in step # 5, the second calculation means 14b of the CPU 14 excludes the influence of disturbances from vibration sources other than the object to be inspected, such as shock vibration when the conveyor is stopped and shock vibration when the air cylinder is operating. N2 from the larger peak values arranged in the above-mentioned order stored in the RAM 17 for each frequency component. _IP Excluding 2 data, eg 2 data, N _IP +1 is the maximum effective peak value P _EM1 ... P _EMn Is extracted and stored in the RAM 17.
[0089]
Further, the second calculation means 14b moves to step # 6, and the second sound detector 201 out of the maximum peak values obtained one by one for each sampling period within one inspection time. The maximum peak value of the amplitude of the sound outside the detected soundproof room ST2 is searched.
[0090]
Then, the second calculating means 14b determines whether or not the maximum peak value exceeds a preset threshold value, and detects the maximum peak value when the maximum peak value exceeds the threshold value. Data of the sound amplitude of the object ST1 sampled within the sampling period is deleted.
[0091]
In this case, the second calculation means 14b deletes the sound of the inspected object ST1 within the sampling period including the time when the sound detected by the second sound detector 201 becomes larger than the threshold value. The sound of the inspected object ST1 within the previous one or two or more sampling periods is deleted.
[0092]
The second computing means 14b deletes the sound of the subject ST1 within the sampling period including the time when the sound detected by the second sound detector 201 becomes smaller than the threshold value, and thereafter The sound of the inspected object ST1 within one or more sampling periods is deleted.
[0093]
For example, if a carriage passes near the soundproof room ST2 that houses the object ST1 or an object falls in the vicinity of the soundproof room ST2, the second sound detector 201 detects when the truck passes. A sound or a sound when an object falls is detected, and an electric signal SA2 corresponding to the sound waveform is output.
[0094]
FIG. 7 is a waveform diagram of the electrical signal SA2 when the second sound detector 201 detects the passing of the carriage or the falling of the object. In this electrical signal SAn + 1, a waveform Ga appears when a sound when a carriage passes or a sound when an object falls is detected within a sampling period Ta.
[0095]
On the other hand, also in the electric signal SA1 output from the first sound detector 1, the waveform Gb due to the passing sound of the carriage or the falling sound of the object appears.
Therefore, the second calculation means 14b searches for the maximum peak value of the amplitude of the sound outside the soundproof room ST2 detected by the second sound detector 201 within the sampling period Ta, that is, the peak value by the waveform Ga. When the maximum peak value exceeds a preset threshold value, the sound amplitude data of the object ST1 sampled within the sampling period Ta in which the maximum peak value is detected is deleted. That is, the electric signal SA1 within the sampling period Ta is deleted.
[0096]
Next, in step # 7, the third computing means 14c of the CPU 14 sets each set value P preset in the RAM 17 for each frequency component. _T1 ... P _Tn And each of these set values P _T1 ... P _Tn Each effective maximum peak value P corresponding to _EM1 ... P _EMn And the effective maximum peak value P _EM1 ... P _EMn If any one of them exceeds the set value, it is determined that “the inspection object 1 is abnormal”.
[0097]
Note that the set values are stored in the RAM 17 in the same manner as described above for the normal inspected objects, for example, N = 100 (pieces) before the actual inspection, and for all the inspected objects. For each frequency component, the effective maximum peak value P is calculated by the following equations (1) and (2) stored in the ROM 18. _EMki Mean value Xi ′ and standard deviation σ _i Is calculated by the CPU 14 and stored in the RAM 17.
[0098]
[Expression 1]

[0099]
Here, i represents a frequency component, i = 1... N. K is the number of test objects, k =... 1N, P _EMki Is the peak value for the frequency component i of the kth object. Mean value Xi ′ and standard deviation σ obtained by the above formulas (1) and (2) _i Based on the following equation (3) stored in the ROM 18 for each frequency component, the set value P _T1 ... P _Tn Ask for.
[0100]
[Expression 2]

[0101]
(Where i = 1 ... n)
Here, Z is a variable that can be arbitrarily set according to the inspection environment and inspection standards.
Above setting value P _T1 ... P _Tn , Effective maximum peak value P _EM1 ... P _EMn And the determination result about the presence or absence of abnormality is displayed by CRT28 for every frequency component in step # 7 as shown in FIG.8 and FIG.9.
[0102]
In these figures, the number n of frequency components is 28. A program for display on the CRT 28 is stored in the ROM 18 in advance.
FIG. 8 shows a CRT display of the inspection result when the waveform is caused by a so-called “hit abnormality” when the rotating portion of the inspection object 1 comes into contact with the fixed portion as shown in FIG.
[0103]
In this figure, the shaded portion of the bar graph indicates an abnormal portion, and it is determined as “abnormal” when the effective maximum peak value is larger than the set value. A band representing an abnormality is displayed by a rectangle 39... Directly above the bar graph indicating the abnormality.
[0104]
As described above, in the first embodiment, a sound external to the device under test 1 is detected by the second sound detector 201, and the detected sound amplitude is a preset threshold value. If it exceeds, data of the amplitude of the sound of the inspection object 1 being sampled is deleted at least, so that it is possible to reliably prevent unexpected background noise and vibration without being affected by the sound of the inspection object 1. It is possible to detect the abnormality of the inspected object 1 reliably.
(2) Next, a second embodiment of the present invention will be described. The same parts as those in FIG. 1 are denoted by the same reference numerals, and detailed description thereof is omitted.
[0105]
In the second embodiment, the abnormality inspection apparatus shown in FIG. 1 is applied to an inspected object ST0 shown in FIG. Hereinafter, description will be given according to the abnormality inspection apparatus shown in FIG.
An object to be inspected ST10 such as a rotating device is placed on the base ST11 via each anti-vibration rubber ST12. A support column ST13 is erected on the base ST11, and a cylinder ST15 is provided at the tip of the support column ST13 via an anti-vibration rubber ST14.
[0106]
A first sound or vibration detector 1 is provided at the driving tip of the cylinder ST15 via an anti-vibration rubber ST16.
The first sound or vibration detector 1 has a function of detecting the sound or vibration of the inspected object ST10 and converting it into a first electric signal SA1 having a waveform corresponding to the sound or vibration waveform. Yes.
[0107]
The first sound or vibration detector 1 will be described below as the first vibration detector 1.
The output side of the first sound detector 1 is connected via the amplifier 2 to each frequency analysis unit 301 to 30n that analyzes frequency components in different frequency bands.
[0108]
The second sound or vibration detection unit 201 is provided on the base ST11 as shown in FIG. 9, and unforeseen background noise or dark vibration, for example, sound or vibration when a carriage passes or sound when an object is dropped. And a function to detect and vibrate, and convert and output to a second electric signal SA2 having a waveform corresponding to the sound or vibration waveform.
[0109]
The second sound or vibration detection unit 201 will be described below as the second vibration detection unit 201.
The output side of the second vibration detector 201 is connected to the frequency analysis unit 30n + 1 via the amplifier 202.
[0110]
On the other hand, the CPU 14 has a function of determining abnormality of the inspected object ST10, and has functions of a first calculation means 14a, a second calculation means 14b, and a third calculation means 14c.
[0111]
The first computing means 14a converts each vibration data detected by the first vibration detector 1 and the second vibration detector 201 into peak value detection circuits 301 to 30n + 1, a multiplexer 10 and analog-digital conversion. It has a function of taking in through the device 11 and rearranging the maximum peak values obtained one by one for each sampling period within one inspection time in order of size.
[0112]
The second calculating means 14b sets each of the maximum peak values rearranged in order of magnitude by the first calculating means 14a in advance corresponding to the content of the disturbance, and sets each specific peak value as an effective maximum peak value. It has a function of extracting each frequency component.
[0113]
Further, the second calculation means 14b is provided on the base 201 detected by the second vibration detector 201 out of the maximum peak values obtained one by one for each sampling period within one inspection time. When the maximum peak value of the external vibration is searched and the maximum peak value exceeds a preset threshold value, the vibration of the object ST10 sampled within the sampling period in which the maximum peak value is detected is detected. It has a function as data deletion means for deleting data.
[0114]
In this case, the second calculation means 14b deletes the sound of the inspected object ST10 within the sampling period including when the vibration detected by the second vibration detector 201 becomes larger than the threshold, The sound of the object ST10 in the previous one or more sampling periods is deleted.
[0115]
The second calculation means 14b deletes the sound of the object to be inspected ST10 within the sampling period including when the vibration detected by the second vibration detector 201 becomes smaller than the threshold, and thereafter The sound of the object ST10 within one or more sampling periods is deleted.
[0116]
The third calculation means 14c has a function of performing abnormality determination by comparing the effective maximum peak value obtained by the second calculation means 14b with a preset value.
[0117]
Next, the operation of the apparatus configured as described above will be described.
First, as in the above-described embodiment, a calculation program (to be described later) is stored in the ROM 18, and an inspection time (for example, 3 seconds), a sampling period Δt (for example, 0.1 seconds), and a peak value are excluded from the RAM 17. Invalid peak value number N _IP (For example, 2), the number of sampling periods to be deleted before and after the sampling period in which an abnormality is detected is stored.
[0118]
Note that the length of the sampling period Δt is preferably set to be shorter than the influence time due to disturbance, for example, impact vibration when the conveyor is stopped.
The length of the inspection time is determined by the abnormal content of the inspected object ST10 to be detected, for example, the rotational abnormality of the rotating device. Further, the number of sampling periods to be deleted before and after the sampling period in which an abnormality is detected is appropriately determined according to the external state.
[0119]
The number of invalid peak values N _IP The number of disturbances is determined based on the number of disturbances empirically obtained within the inspection time and the number of disturbances obtained empirically.
Thus, when the inspected object ST10 such as a rotating device is activated, the first vibration detector 1 detects the vibration of the rotating device and outputs a first electric signal SA1 having a waveform corresponding to the vibration waveform. .
[0120]
The first electric signal SA1 is amplified by the amplifier 2 and then input to the band-pass filters 401 to 40n. The band-pass filters 401 to 40n convert the first electric signals SA1 into signals SC1 to SCn of frequency components in predetermined different frequency bands. Converted.
[0121]
Among these, for example, the signal SCn has a sinusoidal waveform as shown in FIG. The waveforms of the other signals SC1 to SC (n-1) are almost sinusoidal.
[0122]
These signals SC1 to SCn are input to sample and hold circuits 601 to 60n and comparators 701 to 70n, respectively.
Among them, the sample and hold circuit 60n outputs a signal SDn indicating a peak value to the comparator 70n as shown in FIG.
[0123]
The comparator 70n receives the signal SDn from the sample and hold circuit 60n as a set value and also receives the signal SCn from the bandpass filter 40n. When the signal SDn is smaller than the signal SCn, the comparator 70n indicates “0”. SEn is output to the AND circuit 80n, and when the signal SDn is larger than the signal SCn, the signal SEn indicating "1" is output to the AND circuit 80n.
[0124]
Since the signal SCn shown in FIG. 4 monotonously increases in the interval 30, for example, the signal SEn indicating “0” is output to the AND circuit 80n.
On the other hand, since the signal SFn from the decoder 9 that always indicates “1” is output to the other input terminal of the AND circuit 80n except at the time of reset described later, the signal that indicates “0” is output from the AND circuit 80n. SGn is output to the sample and hold circuit 60n.
[0125]
The sample and hold circuit 60n samples the peak value when the signal SFn from the AND circuit 80n is “0”, and holds the peak value when the signal SFn is “1”.
[0126]
Therefore, for example, in the section 30, since the sample and hold circuit 60n is in the peak value sampling state, the signal SDn also increases as shown in FIG. However, since the signal SEn becomes “1” at the inflection point 31 of the signal SCn, the AND circuit 80n outputs the signal SGn indicating “1” to the sample-and-hold circuit 60n, thereby the inflection point 31. Peak value Pn at ₁ Is held, and this peak value Pn ₁ Is output to the multiplexer 10.
[0127]
The peak value P is similarly obtained from the other frequency analysis units 301 to 30 (n−1). ₁₁ ... P _{(n-1) 1} Are output to the multiplexer 10 simultaneously.
[0128]
On the other hand, the second vibration detector 201 detects external vibration on the base ST11 and outputs a second electric signal SA2 having a waveform corresponding to the vibration waveform.
The second electric signal SA2 is amplified by the amplifier 202, and then input to the band pass filter 40n + 1 to be converted into a signal SCn + 1.
[0129]
This signal SCn + 1 is input to the sample and hold circuit 60n + 1 and the comparator 70n + 1.
Among these, the sample and hold circuit 60n + 1 outputs a signal SDn + 1 indicating a peak value to the comparator 70n + 1.
[0130]
The comparator 70n + 1 inputs the signal SDn + 1 from the sample and hold circuit 60n + 1 as a set value and also receives the signal SCn + 1 from the bandpass filter 40n + 1, and receives a signal from the signal SCn + 1. When SDn + 1 is small, the signal SEn + 1 indicating "0" is output to the AND circuit 80n + 1, and when the signal SDn + 1 is larger than the signal SCn + 1, the signal SEn + indicating "1" is output. 1 is output to the AND circuit 80n + 1.
[0131]
Since the signal SFn + 1 from the decoder 9 always indicating “1” is output to the other input terminal of the AND circuit 80n + 1 except for the time of reset, the AND circuit 80n + 1 receives from the AND circuit 80n + 1. A signal SGn + 1 indicating "0" is output to the sample and hold circuit 60n + 1.
[0132]
The sample and hold circuit 60n + 1 samples the peak value when the signal SFn + 1 from the AND circuit 80n + 1 is "0", and holds the peak value when the signal SFn + 1 is "1".
[0133]
Accordingly, the sample-and-hold circuit 60n + 1 holds the peak value at the inflection point of the signal SCn + 1 corresponding to the waveform of the external vibration on the base ST11, and the signal SDn + 1 indicating this peak value is sent to the multiplexer 10. Output to.
[0134]
On the other hand, the CPU 14 outputs signals specifying the required frequency analysis units 301 to 30n and 30n + 1 via the input / output interface 16 to the decoder 9 and the multiplexer 10 via the system bus 20.
[0135]
When this signal SH is sent to the multiplexer 10, the multiplexer 10 causes each peak value (for example, the above Pn) of the frequency components that have passed through the bandpass filters 401 to 40n and 40n + 1 as shown in FIG. ₁ ), And only the signals SD1 to SDn and SDn + 1 indicating the peak value of the signal SCn + 1 corresponding to the external vibration waveform are sequentially supplied to the analog-digital converter 11 as the signals SI1 to SIn and SIn + 1. Is output.
[0136]
Time Δt from the time when these signals SI1 to SIn and SIn + 1 are input to the analog-to-digital converter 11 ₁ After the lapse of (several tens of μs), the CPU 14 applies the pulse signal SJ to the time Δt via the input / output interface 16 as shown in FIG. ₂ It outputs to the differentiation circuit 22 every (about 30 ms or less).
[0137]
The differentiating circuit 22 outputs a signal SK obtained by differentiating the signal SJ to the monostable multivibrator 24 and the analog-digital converter 11.
The signals SI1 to SIn and SIn + 1 at the time when the signal SK is input are converted into digital data signals SL1 to SLn and SLn + 1 by the analog-digital converter 11, and these data signals SL1 to SLn are converted. , SLn + 1 is transmitted to the RAM 17 via the input / output interface 16 and the CPU 14, and the peak value Pn is set at a predetermined address of the RAM 17. ₁ Stored as
[0138]
In FIG. 4, the time Δt between the leading edge of the portion indicating “1” of the signal SK and the trailing edge of the portion indicating “0” of the signal SF1. _Three In order to collect data accurately in real time, Shannon's sampling theorem indicates that Δt _Three <1 / 2F _max (However F _max Is set to satisfy the highest frequency).
[0139]
On the other hand, the monostable multivibrator 24 having received the signal SK outputs the signal SM to the decoder 9 as a strobe signal with the trailing edge of the signal SK as a trigger.
The decoder 9 receiving the signal SM outputs signals SF1 to SFn and SFn + 1 indicating "0" for resetting to the AND circuits 801 to 80n and 80n + 1 designated by the signal SH.
[0140]
As shown in FIG. 4, the AND circuits 801 to 80n and 80n + 1, to which the signals SF1 to SFn and SFn + 1 indicating "0" are input, respectively, indicate the signals indicating "0" to the sample and hold circuits 601 to 601, respectively. 60n, 60n + 1, and the peak value rapidly decreases to, for example, the point 33 of the signal SC shown in FIG. 4, and again as described above, the maximum peak value Pn within the next sampling period Δt shown in FIG. ₂ Is held and output to the multiplexer 10.
[0141]
However, Δt> Δt ₂ X (n + 1) is set, and (n + 1) is the number of frequency analysis units.
On the other hand, for example, the peak value Pn is stored in the RAM 17. ₁ Is stored,
[Δt− {Δt ₂ × (n + 1)}] After a time, the signal SH designating the frequency analysis unit 301 is output to the decoder 9 and the multiplexer 10, and the signal SJ is output to the differentiation circuit 22 in the same manner as the frequency analysis unit 30n. , Peak value P ₁₁ Is converted into a digital data signal SL1 by the analog-to-digital converter 11, and this data signal SL1 is transmitted to the RAM 17, where the peak value P is stored at a predetermined address in the RAM 17. ₁₁ Stored as
[0142]
Thereafter, a signal SF1 indicating “0” for resetting is output to the AND circuit 801, and the sample and hold circuit 601 returns to the sampling state again.
In this way, the maximum peak value P held in each of the other frequency analysis units 302 to 30n + 1 within the sampling period Δt. _{twenty one} ... P _{(n + 1) 1} Are sequentially stored in a predetermined address of the RAM 17.
[0143]
Note that the pulse interval of the signal SJ output from the microcomputer 13 is Δt during the period of sampling SL1 to SLn and SLn + 1 corresponding to the signals SL1 to SLn and SLn + 1 shown in FIG. ₂ For example, the pulse interval from the signal SLn to the sampling of the signal SLn + 1 in the next sampling period Δt is [Δt− {Δt ₂ × (n + 1)}].
[0144]
A program for obtaining such a pulse interval of the signal SJ is stored in the ROM 18 in advance. Furthermore, Δt ₂ Time Δt so that x (n + 1) is equal to the sampling period Δt. ₂ May be set.
[0145]
During each sampling period Δt, the peak value is stored in the RAM 17 sequentially for each frequency analysis unit 301 to 30n, 30n + 1, and is repeated m times as shown in FIG. In the RAM 17, the peak value is stored for each frequency analysis unit 301 to 30n, 30n + 1.
[0146]
Accordingly, m pieces of data are stored for each of the frequency analysis units 301 to 30n and 30n + 1.
When the number of data stored in the RAM 17 reaches m × (n + 1) (where m is the number of samplings and (n + 1) is the number of frequency analysis units), the first calculation of the CPU 14 is performed. The means 14a performs an operation of rearranging the peak values in descending order for each of the frequency analysis units 301 to 30n and 30n + 1, and stores the calculation result in the RAM 17 as shown in FIG.
[0147]
Next, the second calculating means 14b of the CPU 14 uses each frequency in order to eliminate the influence of disturbances from vibration sources other than the inspected object ST10, such as impact vibration when the conveyor is stopped and impact vibration when the air cylinder is operating. For each component, N from the largest peak value arranged in the above-mentioned order stored in the RAM 17 _IP Excluding 2 data, eg 2 data, N _IP +1 is the maximum effective peak value P _EM1 ... P _EMn Is extracted and stored in the RAM 17.
[0148]
In addition, the second calculation means 14b is provided on the base ST11 detected by the second sound detector 201 among the maximum peak values obtained one by one for each sampling period within one inspection time. Find the maximum peak value of external vibration of.
[0149]
Then, the second calculating means 14b determines whether or not the maximum peak value exceeds a preset threshold value, and detects the maximum peak value when the maximum peak value exceeds the threshold value. Data on the amplitude of the sound of the object ST10 sampled within the sampling period is deleted.
[0150]
In this case, the second calculation means 14b deletes the sound of the inspected object ST10 within the sampling period including when the vibration detected by the second vibration detector 201 becomes larger than the threshold, The sound of the object ST10 in the previous one or more sampling periods is deleted.
[0151]
For example, when a carriage passes near the base ST11 on which the inspected object ST10 is placed, or an object falls near the base ST11, the second vibration detector 201 vibrates when the carriage passes. Alternatively, vibration when an object falls is detected, and an electric signal SA2 corresponding to the waveform of these vibrations is output.
[0152]
On the other hand, also in the electric signal SA1 output from the first vibration detector 1, a waveform Gb due to the vibration of passing the carriage or the vibration of dropping the object appears.
Accordingly, the second calculation means 14b searches for the maximum peak value of vibration on the base ST11 detected by the second vibration detector 201 within the sampling period Ta, and this maximum peak value is set in advance. When the value is exceeded, the vibration data of the inspected object ST10 sampled within the sampling period in which the maximum peak value is detected is deleted.
[0153]
Next, the third calculation means 14c of the CPU 14 sets each set value P preset in the RAM 17 for each frequency component. _T1 ... P _Tn And each of these set values P _T1 ... P _Tn Each effective maximum peak value P corresponding to _EM1 ... P _EMn And the effective maximum peak value P _EM1 ... P _EMn If any one of them exceeds the set value, it is determined that “the test object ST10 is abnormal”.
[0154]
As described above, in the second embodiment, the second vibration detector 201 detects a sound external to the object ST10, and the detected vibration exceeds a preset threshold value. Since at least the vibration data of the sample ST10 being sampled is deleted, it is possible to reliably detect unexpected impulsive background noise and dark vibration without being affected by the vibration of the sample ST10. The abnormality determination of the inspected object ST10 can be reliably performed.
(3) Next, a third embodiment of the present invention will be described. The description of the same parts as those in the first embodiment is omitted.
[0155]
The abnormality inspection method of the present invention detects the sound or mechanical vibration of the object to be inspected, divides this sound or mechanical vibration into each frequency component, extracts the maximum peak value in these components, and increases these maximum peak values. In the abnormality inspection method for rearranging in the order of, and determining the abnormality of the inspected object based on the specific maximum peak value among them,
As the specific maximum peak value, the maximum peak value among samples selected at a predetermined ratio with respect to the total number of samples in each frequency component is used.
[0156]
Here, the predetermined ratio is set for each frequency component.
More specifically, in the first embodiment, the first calculation means 14a performs an operation for rearranging the peak values in the descending order for each of the frequency analysis units 301 to 30n and 30n + 1. As shown in FIG.
[0157]
Subsequently, the second calculation means 14b is provided for each frequency component in order to exclude the influence of disturbances from vibration sources other than the object to be inspected, such as impact vibration when the conveyor is stopped and impact vibration when the air cylinder is operating. Are stored in the RAM 17 in order from the largest peak value arranged in the above-mentioned order. _IP Excluding 2 data, eg 2 data, N _IP +1 is the maximum effective peak value P _EM1 ... P _EMn Is extracted and stored in the RAM 17.
[0158]
This second computing means 14b is located outside the soundproof room ST2 detected by the second sound detector 201 among the maximum peak values obtained one by one for each sampling period within one inspection time. Find the maximum peak value of the sound amplitude.
[0159]
Then, the second calculation means 14b determines whether or not the maximum peak value exceeds a preset threshold value, and detects the maximum peak value when the maximum peak value exceeds the threshold value. The data of the amplitude of the sound of the subject ST1 sampled within the sampling period is deleted.
[0160]
In this case, the second calculation means 14b deletes the sound of the inspected object ST1 within the sampling period including the time when the sound detected by the second sound detector 201 becomes larger than the threshold value. The sound of the inspected object ST1 within the previous one or two or more sampling periods is deleted.
[0161]
The second computing means 14b deletes the sound of the object ST1 within the sampling period including the time when the sound detected by the second sound detector 201 becomes smaller than the threshold value, and thereafter The sound of the inspected object ST1 within one or more sampling periods is deleted.
[0162]
On the other hand, in the third embodiment, the sound or mechanical vibration of an object to be inspected such as a rotating device is detected, and the maximum peak value in these components is extracted by dividing this sound or mechanical vibration into each frequency component. Then, these maximum peak values are rearranged in order of magnitude to form a data string similar to that in FIG.
[0163]
If the soundproofing room is sufficiently soundproofed or on a test object that is sufficiently vibration-insulated, rarely generated shock waves or sudden background noise or vibration is larger than the sound or vibration of the test object.
[0164]
Therefore, the data is deleted from the larger one in a predetermined ratio with respect to the total number of samples in the data string, and the abnormality of the object to be inspected is determined based on the maximum peak value in the remaining data. That is, the number of excluded data obtained by the following equation is deleted from the largest one to obtain an effective maximum peak value that is not affected by impact noise or sudden background noise or background vibration. Here, as the designated value, the number of excluded data is the ratio (percentage) of the total number of samples.
[0165]
Number of excluded data = total number of samples− (total number of samples × specified value) / 100 (4)
The predetermined ratio used to determine the data to be excluded is recorded in advance on, for example, a rewritable recording medium, and the number of data actually excluded is calculated by calling the ratio from the recording medium at the time of inspection. Then, measurement data processing is performed using this value, and abnormality determination is performed using the remaining data after exclusion.
[0166]
This predetermined ratio can be set for each frequency component. As this predetermined ratio, for example, the amount to be excluded is set to about 30% or less of the total number of samples.
The specified value is stored in a rewritable recording medium, the calculation is performed once immediately after the execution of the program is started, and the number of excluded data for each frequency is recorded on the RAM.
[0167]
Thereafter, every time inspection is performed, an effective maximum peak value is obtained using the number of excluded data recorded on the RAM, and an abnormality is determined by comparing this with a preset criterion.
As described above, according to the third embodiment, the sound or mechanical vibration of the object to be inspected such as a rotating device is detected, and the sound or the mechanical vibration is separated into each frequency component, and the maximum peak value in these components is detected. Are extracted and rearranged in order of magnitude to create a data string, and the data is deleted from the larger one at a predetermined rate with respect to the total number of samples in this data string, and the maximum peak value in the remaining data Since the abnormality determination of the object to be inspected is performed based on the above, it is needless to say that the same effect as that of the first embodiment is obtained, and furthermore, the decay time of the abnormal sound or abnormal vibration is different for each frequency component. So, if the order of identification (specific number) is matched in the low frequency range where the decay time is long as before, data may be reduced abnormally in the high frequency range where the decay time is short, while Since background noise and dark vibration between 衰時 is matched with a short high-frequency range is not sufficiently attenuated by the low frequency region is sometimes unnecessary data remains. Both of these cause misjudgment, but according to the third embodiment, since such misjudgment can be prevented, more reliable abnormality detection is possible.
[0168]
In addition, this invention is not limited to the said 1st thru | or 3rd embodiment, You may deform | transform as follows.
For example, the abnormality inspection apparatus is not limited to a rotating device as an object to be inspected, and can also be used for abnormality detection of, for example, a nuclear reactor and a manufacturing facility. At this time, the sampling period Δt and the number of invalid peak values N according to the length of the influence time of the disturbance _IP Need to be changed accordingly.
[0169]
Furthermore, although a CRT is used as the display unit, it may be a printer.
Each of the peak detection circuits 501 to 50n + 1 may be an effective value detection circuit, or may not only hold a positive half wave peak value but also hold a negative half wave peak value to obtain an absolute value peak value. It may be a detection circuit. Further, the peak detection circuits 501 to 50n + are set so as to be in a hold state when an impact or sudden background noise or background vibration is input using the output signal of the second sound or vibration detector 201. Each sample-and-hold control signal 1 may be directly controlled.
[0170]
FIG. 10 is a configuration diagram in a case where a hold state is entered when such shocking or sudden background noise or dark vibration enters. The output side of the second sound or vibration detector 201 is connected to each frequency analysis unit 50n + 1 via an amplifier 202, a band pass filter 203, a peak value or effective value detection circuit 204, and a comparison circuit 205. .
[0171]
The frequency analyzer 50n + 1 is connected between the sample and hold circuit 60n + 1, the comparator 70n + 1, the AND circuit 80n + 1, and the comparator 70n + 1 and the AND circuit 80n + 1. And AND circuit 90n + 1.
[0172]
Among these, the second sound or the electric signal from the vibration detector 201 is inputted to the input part of the AND circuit 90n + 1.
In the first to third embodiments, the sound or mechanical vibration of the object to be inspected is detected, the sound or mechanical vibration is divided into frequency components, and the maximum peak value in these components is extracted. The maximum peak values are rearranged in order of magnitude, and the case where the abnormality determination of the inspected object is performed based on the predetermined specific maximum peak value has been described. This is because the sound or mechanical vibration of the inspected object is detected. Even when this sound or mechanical vibration is divided into frequency components and the maximum peak value in these components is extracted to determine abnormality of the inspected object, external sound or mechanical If vibration is detected and the detected external sound or mechanical vibration exceeds a preset threshold, at least the sound or mechanical vibration of the object to be inspected during sampling is deleted to ensure reliability. High anomaly detection Kill.
[0173]
【The invention's effect】
As detailed above According to the present invention, the sampling in which the sound or mechanical vibration of the inspection object detected by the first sound or vibration detector is deleted within the sampling period in which the maximum peak value is detected, and at least the maximum peak value is detected. Since the sound or mechanical vibration of the inspected object detected by the first sound or vibration detector within one or more sampling periods after the period is deleted, A highly reliable abnormality inspection method that can reliably determine the abnormality of an object to be inspected by preventing erroneous determination due to unexpected impact or sudden background noise or vibration. And its device Can provide.
[Brief description of the drawings]
FIG. 1 is a configuration diagram showing a first embodiment of an abnormality inspection apparatus according to the present invention.
FIG. 2 is a layout view of first and second sound or vibration detectors.
FIG. 3 is a flowchart of an abnormality detection operation.
FIG. 4 is a timing chart of an abnormality detection operation.
FIG. 5 is a diagram illustrating a maximum peak value that is sequentially input to an arithmetic control unit of the apparatus.
FIG. 6 is a schematic diagram of maximum peak values rearranged in descending order in the RAM of the calculation control unit of the apparatus.
FIG. 7 is a diagram showing signal waveforms of a sound of an object to be inspected and its external sound.
FIG. 8 is a diagram showing a display example of inspection results.
FIG. 9 is a configuration diagram showing a second embodiment of an abnormality inspection apparatus according to the present invention.
FIG. 10 is a configuration diagram showing a modification of the device of the present invention.
[Explanation of symbols]
1 ... first sound or vibration detector,
ST1 ... Inspection object,
301 to 30n, 30n + 1 ... frequency analysis unit,
201 ... 2nd sound or vibration detection part,
501-50n, 50n + 1 ... peak value detection circuit,
10: Multiplexer,
13 ... Microcomputer,
14a ... 1st calculating means,
14b ... second calculation means,
14c ... 3rd calculating means.

Claims (7)

Sound or mechanical vibration of the object to be inspected, which is housed in soundproof chamber detected by the first sound or vibration detector disposed in the soundproofing chamber, these the sound or divided the mechanical vibration to each frequency component In the abnormality inspection method of extracting each maximum peak value in the frequency component , rearranging these maximum peak values in order of magnitude, and performing abnormality determination of the inspected object based on a predetermined specific maximum peak value,
The detected by a second sound or vibration detector disposed outside sound or mechanical vibration to the soundproofing outdoor against the object to be inspected, the said sound or said mechanical vibration of said detected external When the maximum peak value exceeds a preset threshold value, the sound of the object to be inspected or the machine detected by the first sound or vibration detector within the sampling period in which the maximum peak value is detected remove the vibration,
The sound or the mechanical vibration of the object to be inspected detected by the first sound or vibration detector within at least one sampling period after the sampling period in which the maximum peak value is detected is deleted. An abnormality inspection method characterized by: Deleting the sound or the mechanical vibration of the object detected by the first sound or vibration detector within one or more sampling periods before the sampling period in which the maximum peak value is detected. The abnormality inspection method according to claim 1 . The sound or mechanical vibration of the object to be inspected housed in the soundproof room is detected by the first sound or vibration detector arranged in the soundproof room, and the sound or the mechanical vibration is divided into frequency components. In the abnormality inspection apparatus that extracts each maximum peak value in the frequency component, rearranges these maximum peak values in order of magnitude, and performs abnormality determination of the inspected object based on a predetermined specific maximum peak value,
A second sound or vibration detector disposed outside the soundproof room and detecting external sound or mechanical vibration with respect to the object to be inspected;
When the maximum peak value of the external sound or mechanical vibration detected by the second sound or vibration detector exceeds a preset threshold value, a sampling period in which the maximum peak value is detected 1 or 2 or more after the sampling period in which the sound or the mechanical vibration of the inspection object detected by the first sound or vibration detector is deleted and at least the maximum peak value is detected Data deleting means for deleting the sound or the mechanical vibration of the inspection object detected by the first sound or vibration detector within a sampling period;
Abnormal inspection device characterized by comprising a. The data deletion means is configured to detect the sound of the object to be inspected or the machine detected by the first sound or vibration detector within one or more sampling periods before the sampling period in which the maximum peak value is detected. The abnormality inspection apparatus according to claim 3, wherein the mechanical vibration is deleted . The abnormality inspection apparatus according to claim 3, wherein the second sound or vibration detector is arranged outside the soundproof room or on the inspection object via vibration insulation . The object to be inspected is a rotating device housed in the soundproof room via a vibration-proof rubber,
The first sound or vibration detector is disposed in the soundproof room, detects the sound or vibration of the object to be inspected,
The second sound or vibration detector is disposed outside the soundproof room and detects unforeseen background noise or background vibration,
The abnormality inspection apparatus according to claim 3 . A base on which the object to be inspected is placed via an anti-vibration rubber;
A support column erected on the base;
A cylinder provided via a vibration-proof rubber at the tip of the column;
Have
The object to be inspected is a rotating device,
The first sound or vibration detector is provided at the driving tip of the cylinder via an anti-vibration rubber, and detects the sound or vibration of the object to be inspected.
The second sound or vibration detector is provided on the base and detects unforeseen background noise or background vibration,
The abnormality inspection apparatus according to claim 3 .

Images