JP3710126B2 - Measurement waveform diagnosis method and apparatus - Google Patents
Measurement waveform diagnosis method and apparatus Download PDFInfo
- Publication number
- JP3710126B2 JP3710126B2 JP2001241036A JP2001241036A JP3710126B2 JP 3710126 B2 JP3710126 B2 JP 3710126B2 JP 2001241036 A JP2001241036 A JP 2001241036A JP 2001241036 A JP2001241036 A JP 2001241036A JP 3710126 B2 JP3710126 B2 JP 3710126B2
- Authority
- JP
- Japan
- Prior art keywords
- linear prediction
- order
- sound
- prediction coefficient
- measurement
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Images
Landscapes
- Investigating Or Analyzing Materials By The Use Of Ultrasonic Waves (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
- Ultra Sonic Daignosis Equipment (AREA)
Description
【０００１】
【発明の属する技術分野】
本発明は、測定波形診断方法およびその装置に係り、特に、短時間に衝撃音を発生する機械の故障等に起因する音や振動によって得られた計測波形の異常診断を行う測定波形診断方法および装置に関する。
【０００２】
【従来の技術】
従来、機械の故障を診断する方法の一つに、機械から生ずる音や振動等の波形を解析して診断を行うものがある。通常、これらの診断の方法は、信号の計測・信号の特徴解析・良否診断という手順で行なわれている。従来、信号の特徴を解析する方法として、信号のパワーの大小を解析する方法、ＦＦＴ（高速フェリー変換）等を用いて信号のスペクトルを解析する方法、ソナグラム（声紋）を解析する方法、線形予測係数（反射係数）を解析する方法等があった。
【０００３】
信号のパワーを特徴量とする方法は、故障の程度が大きくならなければ故障を特定できなかったり、特定できない故障がある等の理由で適用に限界されたものとなっている。信号のスペクトルを特徴とする方法は、信号のパワーを特徴とする方法と比較すると故障の特定性能は向上するが、元来スペクトルは、長時間の平均的な周波数特性を示すもので、時間情報が含まれていないため特徴解析方法としては十分とはいえない。ソナグラムを特徴量とする方法はこの点を解決するもので、時間情報と周波数情報を同時に分析するため、信号のもつ特徴量を完全に把握することができる。したがって、音声認識や話者の特定、艦船の特定等、多くの分野で使用されている。
【０００４】
ところで、機器の故障を自動的に判定しようとすると、ソナグラムの情報量が多いため自動診断には向いていない。例えば、周波数方向が２００個、時間方向が１０個を判定するための指標とすると、特徴量の数は、２０００個となる。そこで、この特徴量の数を合理的に減少させる方法として、線形予測係数を特徴量とする方法が考えられている。
【０００５】
線形予測係数は、数式１のαｋで表されている係数である。
【数１】
【０００６】
数式１は、任意の時刻の信号はそれ以前の信号の加重和として表すことを示しており、自己回帰モデルと呼ばれている。そして、線形予測係数を基に数式２に示すように記号のパワースペクトルを算出することができる。
【数２】
【０００７】
数式２に示されるようにスペクトルと線形予測係数に対応関係があるため、スペクトルに違いがみられれば線形予測係数に違いがみられるため、線形予測係数を信号の特徴量として利用できることが分かる。また、通常は少数の線形予測係数でスペクトルが計算できるため、特徴量の数が減少する。通常、信号の特徴を識別するためには線形予測係数の個数は２０個程度で十分な場合が多く、ソナグラムと比較すると特徴量の数は減少し、自己診断が可能なレベルまでになり、波形を基にした自動診断において有効な方法となっている。
【０００８】
線形予測係数を算出する方法にはいくつかあるが、これらの予測係数の中で、最大エントロピー法（ＭＥＭ）により求めた線形予測係数は次の特徴があり、故障診断には適している。
１、モデル次数（ｍ）を変えても係数値は変化しない。他の方法はモデル次数を変えると変わる。
２、線形予測係数を用いてスペクトルを計算する場合、時間的に短いデータに対して高精度のスペクトルが得られる。したがって、速い減少でも特徴量を取得できる。
【０００９】
【発明が解決しようとする課題】
しかしながら、最大エントロピー法による線形予測係数は、信号の振幅に関する情報は０次の係数で表現し、高次の係数で信号の特徴（信号の質的特徴）を表現するものである。そこで、通常では、音質が異なる信号に対しては高次の係数で信号の特徴を把握していたが、振幅の違いはあるけれども音質が違わない場合には異常を特定することができないという欠点がある。例えば、製品検査においていつも発生している音としては同じであるが、異常の違いは発生音が大きいか小さいかの違いである場合が相当する。このように、異常を特定する場合に音質は異ならずに振幅だけが異なるものがあり、かかる場合には、０次の係数しか識別に利用できなかった。したがって、音質などを表している高次の次数の線形予測係数から判断しようとする場合、上記のような異常判定ができない問題があったのである。
【００１０】
本発明は上記従来の問題点に着目し、測定波形診断方法およびその装置に係り、特に、高次の線形予測係数にも信号の振幅情報を反映させ、より多数の線形予測係数で特徴を解析し、短い時間に発生する衝撃音を識別できる測定波形診断方法およびその装置を提供することを目的とする。
【００１１】
【課題を解決するための手段】
上記目的を達成するために、本発明に係る測定波形の診断方法は、機械から生じる音あるいは振動の波形を測定・解析して異常を診断する方法において、信号の大きさの平均エネルギを表す０次の線形予測係数と、周波数特性を表す一次以降の線形予測係数とを求めるとともに、前記一次以降の線形予測係数の値に対して前記０次の線形予測係数又はその平方根の値を乗算し、その得られた線形予測係数の乗算値と、これらの値の時間変化に着目して波形異常を診断することを特徴とする。
【００１２】
本発明に係る測定波形診断装置は、機械から生じる音あるいは振動の波形を測定・解析して異常を診断する装置において、音あるいは振動を測定する測定手段と、測定手段からの音あるいは振動より線形予測分析を行なうとともに、０次線形予測係数と、１次以降の線形予測係数とを求める周波数解析手段と、０次以降の信号を受けて、前記一次以降の線形予測係数の値に対して０次の線形予測係数又はその平方根の値を乗算する乗算手段と、乗算値を計測時間に沿って表示する表示手段とからなることを特徴とする。
【００１３】
【作用】
上記方法によれば、機械の近傍にマイクロホンに配置された測定手段が、コツコツ音等の０．５〜１．５秒の間に発生する衝撃音である異常音を測定する。この測定された衝撃音は、音圧の大きさに応じた電気信号に変換されて、周波数解析手段に送信され、周波数解析手段は、音あるいは振動の信号を解析し、０次および１次以降の線形予測分析を行なうとともに、各周波数の０次、１次以降の線形予測係数を求める。この求められた各線形予測係数は、乗算手段に送信され、信号の大きさのエネルギーを表す０次の線形予測係数又は平方根などの０次の線形予測係数に演算を施すことによって得られる値と、周波数特性を表す一次以降の線形予測係数とのそれぞれを乗算する。または、信号の大きさの平均振幅を表す０次の線形予測係数又は平方根などの０次の線形予測係数に演算を施すことによって得られる値と、周波数特性を表す一次以降の線形予測係数のそれぞれを乗算する。この求められた乗算値は、計測時間とともに表示手段に送信され、各乗算値を計測時間に沿って表示する。この乗算値の表示は、従来に比べて衝撃音のところで、信号の大きさのエネルギーあるいは平均振幅が大きくなっているため、その衝撃音の個所で前の部分よりも大きくなっている。この乗算値の差が、表示手段の模式図に、より鮮明に異常部として表示されるため、作業者により容易に識別される。これにより、正常および異常が容易に判定される。
【００１４】
【発明の実施の形態】
以下に、本発明に係る測定波形診断方法およびその装置の具体的実施の形態を図面にしたがって詳細に説明する。図１は、本発明に係る測定波形診断装置１の一例を示すブロック図であり、図２は測定する音圧の一例を示す図であり、短時間に生ずる衝撃音の音圧の波形である。以下では、被測定物体１０の機械の音圧について記述するが、機械の振動でも良い。図３は測定した音圧を測定波形診断装置１で求めた表示図の一例である。
【００１５】
図１において、測定波形診断装置１は、検査対象である機械等の被測定物体１０の近傍に、音を収集するマイクロホン等の測定手段１２が配置されている。この測定手段１２は、実施形態では機械の近傍にマイクロホンが用いられて構成されており、図２に示すような測定対象物から周期的に発生される波形Ｐｓを測定している。例えば、図２は車両用椅子１０ａに用いたリニアスライド１０ｂで生ずる３分間の波形一例であり、０．５〜１．５秒の間で部品同士の衝突により発生する異常音である衝撃音（コツコツ音Ｄａ）が多数発生していることが確かめられている。なお、測定手段１２には、振動を測定する場合には振動加速度計を用いて振動波形を抽出するようにすればよい。実施形態では、上記車両用椅子１０ａのように被測定物体１０の機械が駆動したときに生ずる音圧を集音し、音圧の大きさに応じた電気信号に変え、図示しない増幅器、フィルタを介して、周波数解析手段１４に入力している。周波数解析手段１４は、音（あるいは振動）の信号を解析し、０次および１次以降の線形予測係数αｋを求める。この解析された音あるいは振動の０次線形予測係数α０は平均エネルギを示し、１次以降の線形予測係数α１、α２、α３、…は音質を示す。この０次の平均エネルギの線形予測係数α０と、および、音質の１次以降の線形予測係数αｋとが乗算手段１６に送信される。乗算手段１６は、０次の線形予測係数α０と、一次以降の線形予測係数α１、α２、α３、…とのそれぞれを数式３に示すごとく乗算し、この乗算した各乗算値α′ｋと、計測時間Ｔｎを表示手段１８に送信する。周波数解析手段１４および乗算手段１６は、コントローラ等の制御手段２０により構成されている。
【数３】
【００１６】
表示手段１８は、各乗算値α′ｋとその時間変化を、図３に示すように模式図表示する。
または、他の実施例の第１周波数解析手段１４Ａは、音あるいは振動の信号を解析し、０次および１次以降の線形予測係数αｋを求める。この解析された音あるいは振動の０次線形予測係数α０は、平均エネルギーから平方根の平均振幅を求め、１次以降の線形予測係数α１、α２、α３、…は高次の音質を求める。この平均振幅の０次線形予測係数α０の平方根と、および、音質の１次以降の線形予測係数αｋが第１乗算手段１６Ａに送られる。第１乗算手段１６Ａは、０次線形予測係数α０の平方振幅と、一次以降の線形予測係数α１、α２、α３、…とのそれぞれを数式４に示すごとく乗算し、この乗算した各乗算値αk″と、その時間変化を表示手段１８に送信する。
【数４】
【００１７】
表示手段１８は、各乗算値αk″とその時間変化を、図３と同様に、模式図表示する。
表示手段１８は、各乗算値αk′、あるいはαk″と、その時間変化を、図３に示すように模式図に示し、ある単位時間に収集した波形データの中に衝撃的に発生する衝撃音、例えば、コツ、コツ音等が測定されるために、突発的に衝撃音が入った場合でも、従来の表示手段よりも、より鮮明に差異が表示される表示手段１８を見て判別することにより、機械の良否の診断が個人差による判断のばらつきも無くなり、検査ミスも無くなる。また、後述するように、閾値を超える各乗算値αk′、あるいはαk″を測定することにより、正常・異常品の診断を行うこともできる。
【００１８】
上記構成により、次に作動について説明する。先ず、機械の近傍にマイクロホンに配置された測定手段１２は、図２は示すような、車用椅子に用いたリニアスライドで生ずるコツコツ音等の０．５〜１．５秒の間に発生する衝撃音である異常音を測定する。この測定された異常音は、音圧の大きさに応じた電気信号に変換されて、周波数解析手段１４に入力されている。周波数解析手段１４は、音あるいは振動の信号を解析し、０次および１次以降の線形予測係数αｋを求める。この求められた線形予測係数αｋは、乗算手段１６に送信され、数式３、あるいは数式４により乗算されて、乗算値αk′、あるいはαk″が求められる。この求められた乗算値αk′、あるいはαk″は、計測時間Ｔｎとともに表示手段１８に送信され、各乗算値αk′、あるいはαk″と、その時間変化を、後述する図４に示すＭＥＭで求めた模式図と比較して、図３に示すように、異常部Ｕａが模式図により鮮明に表示される。図３は、図４に比べて、乗算値の表示が、従来に比べて衝撃音のところで、信号の大きさのエネルギーあるいは平均振幅が大きくなっているため、その衝撃音の個所の異常部Ｕａが前の部分よりも大きくなっている。この乗算値の差が、表示手段１８の模式図に、より鮮明に異常部Ｕａとして表示されるため、作業者により容易に識別される。これにより、正常および異常が容易に判定される。
【００１９】
次に、比較例として、ＭＥＭにより線形予測係数で求めた模式図を図４に示す。
上記の故障診断方法では、信号の大きさのエネルギーを表す０次の線形予測係数と、一次以降の線形予測係数とのそれぞれを乗算し、その得られた乗算値の線形予測係数の特徴量と、計測時間との相関を模式図にして故障を診断するために、高次の線形予測係数にも信号の振幅情報を反映させ、より多数の線形予測係数で特徴を解析し、短い時間に発生する衝撃音をより正確に、精度良く識別できる。
【００２０】
上記のように計測信号から線形予測係数αk’（またはαk”以下同じ）を求めるが、図３のようにモニタ表示して視覚判断するようにしてもよいが、このようにして求めた線形予測係数αk’をニューラルネットワークなどに入力し、自動判別するようにできる。特徴量が顕著に表されるように式３，４の如くデータ処理しているため、自動判定が的確に行われる。
【００２１】
【発明の効果】
以上説明したように本発明によれば、測定波形診断方法およびその装置において、信号の大きさの平均エネルギを表す０次の線形予測係数と、周波数特性を表す一次以降の線形予測係数とを求めるとともに、前記一次以降の線形予測係数の値に対して前記０次の線形予測係数又はその平方根の値を乗算し、その得られた線形予測係数の乗算値と、これらの値の時間変化に着目して波形異常を診断する。これにより、機械音等の波形の時間的特徴、周波数的特徴、その大小を効果的に解析できるので、機械の故障等により生ずる波形の特徴を模式図に、より鮮明に、精度良く表すことができ、この模式図が作業者により識別されることにより、正常および異常が容易に判定される。
【図面の簡単な説明】
【図１】 本発明に係る測定波形診断装置の一例を示すブロック図である。
【図２】 本発明に係る測定波形診断装置により、測定する音圧の一例を示す図である。
【図３】 本発明に係る測定波形診断装置により求めた音圧を模式図にした例を示す図である。
【図４】 従来のＭＥＭにより求めた音圧を模式図にした例を示す図である。
【符号の説明】
１………測定波形診断装置、１０………被測定物体、１２………測定手段、
１４………周波数解析手段、１６………乗算手段、１８………表示手段、
２０………制御手段。[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a measurement waveform diagnosis method and apparatus, and more particularly to a measurement waveform diagnosis method for diagnosing abnormality of a measurement waveform obtained by sound and vibration caused by a failure of a machine that generates an impact sound in a short time, and the like. Relates to the device.
[0002]
[Prior art]
Conventionally, as one of the methods for diagnosing a machine failure, there is a method for diagnosing by analyzing waveforms such as sound and vibration generated from the machine. Usually, these diagnosis methods are performed by procedures of signal measurement, signal feature analysis, and pass / fail diagnosis. Conventionally, as a method of analyzing signal characteristics, a method of analyzing the power level of a signal, a method of analyzing a signal spectrum using FFT (Fast Ferry Transform), a method of analyzing a sonagram (voiceprint), a linear prediction There was a method of analyzing the coefficient (reflection coefficient).
[0003]
The method using the signal power as a feature amount is limited to application because the failure cannot be specified unless the failure level becomes large, or there is a failure that cannot be specified. The method characterized by the spectrum of the signal improves the specific performance of the fault compared to the method characterized by the power of the signal, but the spectrum originally shows an average frequency characteristic over a long period of time. Is not sufficient as a feature analysis method. The method using a sonagram as a feature amount solves this point. Since the time information and the frequency information are analyzed simultaneously, the feature amount of the signal can be completely grasped. Therefore, it is used in many fields such as voice recognition, speaker identification, and ship identification.
[0004]
By the way, an attempt to automatically determine a device failure is not suitable for automatic diagnosis due to the large amount of information in the sonogram. For example, if it is an index for determining 200 in the frequency direction and 10 in the time direction, the number of feature amounts is 2000. Therefore, as a method for rationally reducing the number of feature quantities, a method using linear prediction coefficients as feature quantities is considered.
[0005]
The linear prediction coefficient is a coefficient represented by α k in Equation 1.
[Expression 1]
[0006]
Formula 1 indicates that a signal at an arbitrary time is expressed as a weighted sum of signals before that, and is called an autoregressive model. Based on the linear prediction coefficient, the power spectrum of the symbol can be calculated as shown in Equation 2.
[Expression 2]
[0007]
As shown in Equation 2, since the spectrum and the linear prediction coefficient have a correspondence relationship, if the spectrum is different, the linear prediction coefficient is different, so that the linear prediction coefficient can be used as the signal feature amount. In addition, since the spectrum can be usually calculated with a small number of linear prediction coefficients, the number of feature amounts is reduced. Usually, about 20 linear prediction coefficients are sufficient to identify signal features. Compared with sonagram, the number of features is reduced to a level where self-diagnosis is possible. It is an effective method in automatic diagnosis based on
[0008]
There are several methods for calculating the linear prediction coefficient. Among these prediction coefficients, the linear prediction coefficient obtained by the maximum entropy method (MEM) has the following characteristics and is suitable for fault diagnosis.
1. The coefficient value does not change even if the model order (m) is changed. Other methods change with changing model order.
2. When calculating a spectrum using a linear prediction coefficient, a highly accurate spectrum can be obtained for temporally short data. Therefore, the feature amount can be acquired even with a rapid decrease.
[0009]
[Problems to be solved by the invention]
However, in the linear prediction coefficient based on the maximum entropy method, information related to the amplitude of the signal is expressed by a zeroth-order coefficient, and a signal characteristic (qualitative characteristic of the signal) is expressed by a higher-order coefficient. Therefore, normally, for signals with different sound quality, the characteristics of the signal have been grasped with higher-order coefficients, but there is a drawback in that if there is a difference in amplitude but the sound quality is not different, an abnormality cannot be identified. There is. For example, the sound that is always generated in the product inspection is the same, but the difference in abnormality corresponds to the difference in whether the generated sound is large or small. As described above, when an abnormality is specified, there is a sound quality that is not different but only an amplitude is different. In such a case, only the zeroth order coefficient can be used for identification. Therefore, when trying to determine from higher-order linear prediction coefficients representing sound quality and the like, there is a problem that the above-described abnormality determination cannot be performed.
[0010]
The present invention pays attention to the above-mentioned conventional problems, and relates to a measurement waveform diagnosis method and apparatus, and in particular, reflects signal amplitude information in higher-order linear prediction coefficients and analyzes features with a larger number of linear prediction coefficients. An object of the present invention is to provide a measurement waveform diagnostic method and apparatus capable of identifying a shock sound generated in a short time.
[0011]
[Means for Solving the Problems]
In order to achieve the above object, a method for diagnosing a measured waveform according to the present invention is a method for diagnosing an abnormality by measuring and analyzing a sound or vibration waveform generated from a machine. Obtaining the next linear prediction coefficient and the first and subsequent linear prediction coefficients representing frequency characteristics, and multiplying the value of the first and subsequent linear prediction coefficients by the zeroth order linear prediction coefficient or the value of the square root thereof; The waveform abnormality is diagnosed by paying attention to the multiplication value of the obtained linear prediction coefficient and the time change of these values.
[0012]
The measurement waveform diagnosis apparatus according to the present invention is a device that measures and analyzes a sound or vibration waveform generated from a machine and diagnoses an abnormality. The measurement waveform diagnosis apparatus is linear from a sound or vibration from a measurement means that measures sound or vibration. Frequency analysis means for performing a prediction analysis and obtaining a 0th-order linear prediction coefficient and a linear prediction coefficient after the first order, and receiving a signal after the 0th order, and 0 for the value of the linear prediction coefficient after the first order. It is characterized by comprising multiplication means for multiplying the next linear prediction coefficient or a square root value thereof, and display means for displaying the multiplication value along the measurement time.
[0013]
[Action]
According to the above method, the measurement means arranged in the microphone in the vicinity of the machine measures the abnormal sound that is an impact sound generated during 0.5 to 1.5 seconds, such as a bang sound. The measured impact sound is converted into an electric signal corresponding to the magnitude of the sound pressure and transmitted to the frequency analysis means. The frequency analysis means analyzes the sound or vibration signal, and performs the 0th order and the 1st order and thereafter. And the linear prediction coefficients of the 0th order and the 1st order of each frequency are obtained. Each of the obtained linear prediction coefficients is transmitted to the multiplying means, and a value obtained by performing an operation on a zeroth-order linear prediction coefficient representing a signal magnitude energy or a zeroth-order linear prediction coefficient such as a square root. Multiply each of the first and subsequent linear prediction coefficients representing frequency characteristics. Alternatively, a value obtained by performing an operation on a zeroth-order linear prediction coefficient representing the average amplitude of the signal magnitude or a zeroth-order linear prediction coefficient such as a square root, and a linear prediction coefficient after the first order representing the frequency characteristics, respectively. Multiply The obtained multiplication value is transmitted to the display unit together with the measurement time, and each multiplication value is displayed along the measurement time. The display of this multiplication value is larger than the previous portion at the location of the impact sound because the energy or average amplitude of the signal magnitude is greater at the impact sound than in the conventional case. The difference between the multiplication values is more clearly displayed as an abnormal part in the schematic diagram of the display means, and thus is easily identified by the operator. Thereby, normality and abnormality are easily determined.
[0014]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, specific embodiments of a measurement waveform diagnosis method and apparatus according to the present invention will be described in detail with reference to the drawings. FIG. 1 is a block diagram showing an example of a measurement waveform diagnostic apparatus 1 according to the present invention, and FIG. 2 is a diagram showing an example of a sound pressure to be measured, which is a waveform of a sound pressure of an impact sound generated in a short time. . Hereinafter, although the sound pressure of the machine of the measured
[0015]
In FIG. 1, the measurement waveform diagnostic apparatus 1 includes a
[Equation 3]
[0016]
The display means 18 displays each multiplication value α ′ k and its change over time schematically as shown in FIG.
Alternatively, the first frequency analysis unit 14A of another embodiment analyzes a sound or vibration signal and obtains linear prediction coefficients α k of the 0th order and the 1st order and later. The 0th-order linear prediction coefficient α 0 of the analyzed sound or vibration is obtained by calculating the mean amplitude of the square root from the average energy, and the linear prediction coefficients α 1 , α 2 , α 3 ,. Ask. The square root of the zeroth-order linear prediction coefficient α 0 of the average amplitude and the first and subsequent linear prediction coefficients α k of the sound quality are sent to the first multiplication means 16A. First multiplier means 16A includes a square amplitude of the zero-order linear prediction coefficient alpha 0, the linear prediction coefficients of the primary since α 1, α 2, α 3 , ... , respectively and multiplied as shown in Equation 4, and the multiplication Each multiplication value α k ″ and its time change are transmitted to the display means 18.
[Expression 4]
[0017]
The display means 18 displays each multiplication value αk ″ and its change over time in a schematic diagram as in FIG.
The display means 18 shows each multiplication value αk ′ or αk ″ and its change over time in a schematic diagram as shown in FIG. 3, and an impact sound generated shockingly in the waveform data collected in a certain unit time. For example, even when an impact sound suddenly enters because a knack, a knack sound, or the like is measured, it is determined by looking at the display means 18 that displays the difference more clearly than the conventional display means. This eliminates differences in judgments due to individual differences in machine quality and inspection errors.In addition, as will be described later, normality / abnormality is measured by measuring each multiplication value αk ′ or αk ″ exceeding a threshold value. The product can also be diagnosed.
[0018]
Next, the operation of the above configuration will be described. First, the measuring means 12 disposed on the microphone in the vicinity of the machine is generated for 0.5 to 1.5 seconds such as a knack sound generated by a linear slide used in a car chair as shown in FIG. Measure the abnormal sound that is an impact sound. The measured abnormal sound is converted into an electrical signal corresponding to the magnitude of the sound pressure and input to the frequency analysis means 14. The frequency analysis means 14 analyzes a sound or vibration signal and obtains linear prediction coefficients α k for the zeroth order and the first and subsequent orders. The obtained linear prediction coefficient α k is transmitted to the multiplication means 16 and is multiplied by Equation 3 or Equation 4 to obtain a multiplication value α k ′ or α k ″. The obtained multiplication value α k ′ or α k ″ is transmitted to the display means 18 together with the measurement time Tn, and each multiplication value α k ′ or α k ″ and the time change thereof are schematic diagrams obtained by MEM shown in FIG. 4 to be described later. 3, the abnormal portion Ua is clearly displayed in a schematic diagram as shown in Fig. 3. In Fig. 3, the multiplication value is displayed in an impact sound as compared with Fig. 4. Since the energy of the signal magnitude or the average amplitude is larger, the abnormal portion Ua at the location of the impact sound is larger than the previous portion, and the difference in the multiplication values is shown in the schematic diagram of the display means 18. Because it is displayed more clearly as the abnormal part Ua, Are easily identified by the user. Thus, normal and abnormal are easily determined.
[0019]
Next, as a comparative example, a schematic diagram obtained by a linear prediction coefficient by MEM is shown in FIG.
In the above fault diagnosis method, the zeroth-order linear prediction coefficient representing the energy of the signal magnitude is multiplied by the linear prediction coefficient after the first order, and the feature quantity of the linear prediction coefficient of the obtained multiplication value is obtained. In order to diagnose a failure with a schematic diagram of the correlation with the measurement time, the signal amplitude information is also reflected in the higher-order linear prediction coefficients, and the characteristics are analyzed with a larger number of linear prediction coefficients, and they occur in a short time. Can be identified more accurately and accurately.
[0020]
As described above, the linear prediction coefficient α k ′ (or α k ″ is the same) is obtained from the measurement signal, but it may be displayed visually on the monitor as shown in FIG. The linear prediction coefficient α k ′ can be automatically input by inputting it into a neural network, etc. Since the data is processed as shown in equations 3 and 4 so that the feature quantity is remarkably expressed, automatic determination is performed accurately. Is called.
[0021]
【The invention's effect】
As described above, according to the present invention, in the measurement waveform diagnosis method and apparatus therefor, the zeroth-order linear prediction coefficient representing the average energy of the signal magnitude and the first and subsequent linear prediction coefficients representing the frequency characteristics are obtained. At the same time, the value of the linear prediction coefficient after the first order is multiplied by the linear prediction coefficient of the 0th order or the square root thereof, and attention is paid to the multiplication value of the obtained linear prediction coefficient and the time change of these values. To diagnose waveform abnormalities. As a result, the temporal characteristics, frequency characteristics, and magnitudes of waveforms such as mechanical sounds can be analyzed effectively, so that the characteristics of waveforms caused by machine failures can be represented more clearly and accurately. The schematic diagram is identified by the operator, so that normality and abnormality are easily determined.
[Brief description of the drawings]
FIG. 1 is a block diagram showing an example of a measurement waveform diagnostic apparatus according to the present invention.
FIG. 2 is a diagram showing an example of sound pressure to be measured by the measurement waveform diagnostic apparatus according to the present invention.
FIG. 3 is a diagram showing an example in which the sound pressure obtained by the measurement waveform diagnostic apparatus according to the present invention is schematically shown.
FIG. 4 is a diagram showing an example in which a sound pressure obtained by a conventional MEM is schematically illustrated.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ......... Measurement waveform diagnostic apparatus, 10 ......... Measurement object, 12 ......... Measurement means,
14... Frequency analysis means 16... Multiplication means 18.
20: Control means.
Claims (2)
信号の大きさの平均エネルギを表す０次の線形予測係数と、周波数特性を表す一次以降の線形予測係数とを求めるとともに、前記一次以降の線形予測係数の値に対して前記０次の線形予測係数又はその平方根の値を乗算し、その得られた線形予測係数の乗算値と、これらの値の時間変化に着目して波形異常を診断することを特徴とする測定波形の診断方法。 In the method of diagnosing abnormalities by measuring and analyzing the sound or vibration waveform generated by the machine,
A zeroth-order linear prediction coefficient representing the average energy of the signal magnitude and a first and subsequent linear prediction coefficients representing frequency characteristics are obtained, and the zeroth order linear prediction is performed with respect to the values of the first and subsequent linear prediction coefficients. A method of diagnosing a measured waveform, wherein a waveform abnormality is diagnosed by multiplying a coefficient or a square root value thereof, and paying attention to a multiplication value of the obtained linear prediction coefficient and a temporal change of these values.
音あるいは振動を測定する測定手段と、
測定手段からの音あるいは振動より線形予測分析を行なうとともに、０次線形予測係数と、１次以降の線形予測係数とを求める周波数解析手段と、
０次以降の信号を受けて、前記一次以降の線形予測係数の値に対して０次の線形予測係数又はその平方根の値を乗算する乗算手段と、
乗算値を計測時間に沿って表示する表示手段とからなることを特徴とする測定波形診断装置。 In a device that diagnoses abnormalities by measuring and analyzing the sound or vibration waveform generated by the machine,
Measuring means for measuring sound or vibration;
Frequency analysis means for performing linear prediction analysis based on sound or vibration from the measurement means, and obtaining zero-order linear prediction coefficients and linear prediction coefficients after the first order;
Multiplying means for receiving a 0th order signal and subsequent signals and multiplying the first order and subsequent linear prediction coefficient values by a 0th order linear prediction coefficient or a square root value thereof;
A measurement waveform diagnostic apparatus comprising display means for displaying a multiplication value along a measurement time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2001241036A JP3710126B2 (en) | 2001-08-08 | 2001-08-08 | Measurement waveform diagnosis method and apparatus |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2001241036A JP3710126B2 (en) | 2001-08-08 | 2001-08-08 | Measurement waveform diagnosis method and apparatus |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| JP2003057210A JP2003057210A (en) | 2003-02-26 |
| JP3710126B2 true JP3710126B2 (en) | 2005-10-26 |
Family
ID=19071554
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| JP2001241036A Expired - Fee Related JP3710126B2 (en) | 2001-08-08 | 2001-08-08 | Measurement waveform diagnosis method and apparatus |
Country Status (1)
| Country | Link |
|---|---|
| JP (1) | JP3710126B2 (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2007199017A (en) * | 2006-01-30 | 2007-08-09 | Omron Corp | Method and apparatus for clearing abnormal noise, and abnormal noise inspection apparatus |
| JP2007327884A (en) * | 2006-06-08 | 2007-12-20 | Matsushita Electric Works Ltd | Quality judging device |
| JP2008270948A (en) * | 2007-04-17 | 2008-11-06 | Sharp Corp | Information terminal device, control method of the device, program for control over the device, and recording medium where the program is recorded |
| JP5106329B2 (en) * | 2008-09-13 | 2012-12-26 | 中国電力株式会社 | Equipment abnormality diagnosis method and system |
| US11372093B2 (en) * | 2018-09-14 | 2022-06-28 | Fujifilm Sonosite, Inc. | Automated fault detection and correction in an ultrasound imaging system |
-
2001
- 2001-08-08 JP JP2001241036A patent/JP3710126B2/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| JP2003057210A (en) | 2003-02-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP2156159B1 (en) | System and method for bearing fault detection using stator current noise cancellation | |
| EP2374015B1 (en) | System and method for motor fault detection using stator current noise cancellation | |
| CN105705928B (en) | Abnormal sound diagnostic device | |
| CN101124523A (en) | Diagnosis device for course control system | |
| JP4592123B2 (en) | Unsteady signal analyzer and medium on which unsteady signal analysis program is recorded | |
| JPH08292111A (en) | Belt-tension measuring apparatus | |
| JP5106329B2 (en) | Equipment abnormality diagnosis method and system | |
| KR20100094452A (en) | A method for detection and automatic identification of damage to rolling bearings | |
| De Moura et al. | Applications of detrended-fluctuation analysis to gearbox fault diagnosis | |
| EP2345894B1 (en) | Trending of vibration data taking into account torque effect | |
| CN111964909A (en) | Rolling bearing operation state detection method, fault diagnosis method and system | |
| CN111649886A (en) | Abnormality detection device, rotating machine, abnormality detection method, and program | |
| JP2002090267A (en) | Failure diagnosis method | |
| JP3710126B2 (en) | Measurement waveform diagnosis method and apparatus | |
| JP2010266327A (en) | Facility diagnosis device and facility diagnosis method | |
| JP3020349B2 (en) | Error detection method and device | |
| JP3759881B2 (en) | Process diagnosis monitoring system | |
| JP2002090266A (en) | Remaining life-predicting device | |
| JP2003021555A (en) | Abnormality monitoring equipment | |
| EP1686443A1 (en) | Methods, systems, and computer program products for implementing condition monitoring activities | |
| WO2008139134A2 (en) | Detection of an abnormal signal in a compound sampled signal | |
| JP2004205215A (en) | Sound source diagnosing device | |
| JP4049331B2 (en) | Method and apparatus for evaluating diagnostic object | |
| JPH1020925A (en) | Plant diagnostic device | |
| JP2001209628A (en) | Pattern matching method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| A977 | Report on retrieval |
Free format text: JAPANESE INTERMEDIATE CODE: A971007 Effective date: 20040614 |
|
| A131 | Notification of reasons for refusal |
Free format text: JAPANESE INTERMEDIATE CODE: A131 Effective date: 20040625 |
|
| A521 | Request for written amendment filed |
Free format text: JAPANESE INTERMEDIATE CODE: A523 Effective date: 20040823 |
|
| A131 | Notification of reasons for refusal |
Free format text: JAPANESE INTERMEDIATE CODE: A131 Effective date: 20050420 |
|
| A521 | Request for written amendment filed |
Free format text: JAPANESE INTERMEDIATE CODE: A523 Effective date: 20050517 |
|
| TRDD | Decision of grant or rejection written | ||
| A01 | Written decision to grant a patent or to grant a registration (utility model) |
Free format text: JAPANESE INTERMEDIATE CODE: A01 Effective date: 20050805 |
|
| A61 | First payment of annual fees (during grant procedure) |
Free format text: JAPANESE INTERMEDIATE CODE: A61 Effective date: 20050808 |
|
| R150 | Certificate of patent or registration of utility model |
Free format text: JAPANESE INTERMEDIATE CODE: R150 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20080819 Year of fee payment: 3 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20090819 Year of fee payment: 4 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20090819 Year of fee payment: 4 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20100819 Year of fee payment: 5 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20100819 Year of fee payment: 5 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20110819 Year of fee payment: 6 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20130819 Year of fee payment: 8 |
|
| LAPS | Cancellation because of no payment of annual fees |