JP5476413B2 - Diagnostic method for soundness of rotating machinery - Google Patents

Description

  The present invention relates to a soundness diagnosis method for detecting an abnormality of a machine by measuring and analyzing a radiated sound emitted from rotating machinery.

  Conventionally, in order to save labor and rationalize maintenance, a method for diagnosing the soundness of whether there is an abnormality in the drive part of the rotating machine by measuring and analyzing the radiated sound emitted from the rotating machine, Various devices have been proposed.

  For example, Patent Document 1 is a soundness diagnosis method using sound signals of equipment such as a rotating machine, which measures sound signals generated during operation of the equipment and uses a least square method and a maximum entropy method to determine the inherent self The regression coefficient of the regression model is calculated and stored, and the residual is calculated by applying the regression coefficient to the sound signal generated at the time of inspection and measurement of the equipment, and the normality of the equipment at the time of measurement is determined by comparing the magnitude of the residual A soundness diagnosis method using sound signals of equipment such as a rotating machine that has been determined to be abnormal has been proposed.

  Further, in Patent Document 2 below, in an abnormality monitoring device that monitors the presence or absence of an abnormality to be diagnosed, a sensor that captures a predetermined physical quantity that reflects the state of the diagnosis object and obtains a signal that represents the physical quantity; A diagnosis target abnormality determination unit that determines whether there is an abnormality in the diagnosis target based on reference data derived from a reference signal obtained when the target is in a normal state and a diagnostic signal obtained by the sensor during abnormality monitoring A simulated abnormal state generating unit that gives the sensor a physical quantity that simulates when the diagnosis target is in an abnormal state, the reference data, and a timing at which the sensor captures the physical quantity given from the simulated abnormal state generating unit And a self-abnormality determination unit that determines whether there is an abnormality in the abnormality monitoring device itself based on the simulated abnormal component mixed signal obtained by the sensor. A constant monitoring device has been proposed.

  Further, in Patent Document 3 below, a first time series signal carrying a predetermined physical quantity is obtained from one or a plurality of devices in the normal state that are the same as or similar to the device to be detected, and the first time series is obtained. An inverse filter is configured based on the signal, a second time series signal carrying the predetermined physical quantity is obtained from the detected device, and the residual is obtained by applying the inverse filter to the second time series signal. There has been proposed an abnormality detection method for obtaining a signal and detecting an abnormality of the detected device based on the residual signal.

JP 2009-257862 A JP 2003-21555 A Japanese Unexamined Patent Publication No. 7-43259

  However, it is necessary to measure the sound of rotating machinery during daily inspections and periodic inspections, and to determine and predict the occurrence of rotating machine failures and problems from these measurement results. The occurrence and fluctuation of sound due to machine failure and state change vary depending on the individual machine, such as the type and size of the machine. The soundness diagnosis method does not show an appropriate method for determining whether or not repair is necessary.

  On the other hand, a method for properly grasping the state of the rotating machine by measuring the vibration of the rotating machine and applying this vibration measurement value to a certain evaluation standard has been standardized (JIS B 0906 / ISO 10816-1). ). This standard summarizes general guidelines for measuring and evaluating vibrations measured in non-rotating parts such as bearing housings of rotating machinery, and is expressed by the magnitude of vibration for each model. General evaluation criteria are presented.

  In this standard, a method for measuring the vibration of a rotating machine and evaluating the state of the machine based on this vibration measurement value is shown.However, in the actual site, when vibration measurement and sound measurement are compared, ( 1) In vibration measurement, it may be dangerous because the sensor needs to be in direct contact with the rotating machine, but sound measurement is non-contact and highly safe. (2) Vibration The measured values evaluate abnormalities such as shaft misalignment of rotating machinery and wear of bearings on a single scale called the magnitude of vibration, whereas in the case of sound signals, frequency fluctuations and the degree of random noise contamination Because it is possible to evaluate using multiple indicators as in the above, more detailed health diagnosis is possible. (3) In vibration measurement, evaluation is based on the scale of vibration magnitude, so the measurement position is strict. Whereas sound measurement is required Because sound quality (frequency components, etc.) is emphasized rather than the size of sound, sound measurement has more advantages than vibration measurement because it can handle the measurement position relatively flexibly. It was.

  Therefore, in the soundness diagnosis of a rotating machine, a diagnostic method that can evaluate the state of the rotating machine by applying the sound signal obtained by sound measurement to a certain evaluation criterion rather than the evaluation based on the vibration measurement value obtained by vibration measurement. Was desired.

  Therefore, a main object of the present invention is to provide a method for diagnosing the soundness of a rotating machine by applying sound signals obtained by sound measurement to a predetermined evaluation standard.

In order to solve the above-mentioned problem, the present invention according to claim 1 is a method for diagnosing the soundness of a rotating machine by measuring sound,
Measure vibration and sound during operation for the reference rotating machine in advance, obtain a conversion coefficient that maximizes the correlation between these vibration measurement values and sound measurement values, and then operate for the rotating machine to be diagnosed The sound of the hour is measured, the sound measurement value is converted using the conversion coefficient, and the soundness of the rotating machine is diagnosed by applying the converted sound measurement value to the vibration evaluation standard. A method for diagnosing the health of a rotating machine is provided.

  According to the first aspect of the present invention, vibration and sound during operation are measured for a reference rotating machine in advance, and a conversion coefficient that maximizes the correlation between the vibration measurement value and the sound measurement value is obtained. After that, the sound during operation is measured for the rotating machine to be diagnosed, the sound measurement value is converted using the conversion coefficient, and the converted sound measurement value is applied to the vibration evaluation standard to The health is being diagnosed. For this reason, the soundness of the rotating machine can be diagnosed by applying the sound signal obtained by the sound measurement to a certain evaluation standard (vibration evaluation standard).

  As the vibration evaluation standard, an original standard based on uniquely accumulated data or a standard proposed by various organizations can be used. As described in detail later, JIS B 0906 (ISO 10816-1) It is preferable to use the prescribed evaluation criteria.

As a second aspect of the present invention, vibration and sound during operation are measured in advance for a reference rotating machine, and the vibration measurement value is set as a vibration evaluation value z, and a residual analysis is performed from the sound measurement value by an autoregressive model. A first step of obtaining the residual power x of the sound and the correlation coefficient y in the regression map of the residual using
The sound evaluation value u is set by linear weighting by multiplying the residual power x and the correlation coefficient y by the conversion coefficients a and b by the following equation (6), and the vibration evaluation value by the following equation (7): a second step of determining the conversion coefficients a and b under the condition that the square error Q between z and the sound evaluation value u is minimized;
The sound during operation is measured for the rotating machine to be diagnosed, and the residual power X of the sound and the correlation coefficient Y in the regression map of the residual are obtained from this sound measurement value using the residual analysis by the autoregressive model. A third step of converting the sound evaluation value U into the following equation (8):
The sound diagnosis value of a rotating machine according to claim 1, further comprising a fourth step of diagnosing the soundness of the rotating machine by applying the sound evaluation value U to a vibration evaluation standard.

  The invention according to the second aspect defines a specific procedure of the soundness diagnosis method, and includes the following four steps.

  As a first step, vibration and sound during operation are measured in advance for a reference rotating machine, the vibration measurement value is set as a vibration evaluation value z, and the sound is measured using a residual analysis based on an autoregressive model from the sound measurement value. The residual power x and the correlation coefficient y in the regression map of the residual are obtained. Residual analysis using an autoregressive model compares the similarity between the waveform of the vibration measurement value and the waveform of the sound measurement value using the autoregression model, and the difference is the correlation coefficient in the residual power x and the regression map of the residual. This is a method for quantitative evaluation using y.

  As a second step, a sound evaluation value u obtained by multiplying the residual power x and the correlation coefficient y by the conversion coefficients a and b by the above expression (6) and performing linear weighting is set, and the above expression (7) Thus, the conversion coefficients a and b are determined under the condition that the residual Q between the vibration evaluation value z and the sound evaluation value u is minimized. Since the conversion coefficients a and b are determined on the assumption that the residual Q is minimized, a conversion value correlated with the vibration measurement value is obtained by converting the sound measurement value using the conversion coefficient. Will be able to.

  Then, as a third step, the sound is measured for the rotating machine to be diagnosed, and the residual power X of the sound and the correlation in the regression map of the residual using the residual analysis based on the autoregressive model from this sound measurement value The number Y is obtained and converted into a sound evaluation value U by the above equation (8). Since the sound evaluation value U has a correlation with the vibration measurement value, the sound evaluation value U can be applied to the vibration evaluation standard.

  Thereafter, as a fourth step, the sound evaluation value U is applied to the vibration evaluation standard to diagnose the soundness of the rotating machine. This makes it possible to diagnose the soundness of the rotating machine by applying the sound signal obtained by measuring the sound to a certain vibration evaluation standard.

  According to a third aspect of the present invention, there is provided the rotating machine soundness diagnostic method according to any one of the first and second aspects, wherein the vibration evaluation standard uses an evaluation standard defined in JIS B0906.

  In the third aspect of the invention, by using an evaluation standard defined in JIS B0906 as the vibration evaluation standard, it is possible to qualitatively determine the state of the target rotating machine.

  As the present invention according to claim 4, the soundness diagnosis of the rotating machine according to any one of claims 1 to 3, wherein the reference rotating machine is set in units of several tens of units so that the vibration evaluation value z does not overlap. A method is provided.

  In the invention according to the fourth aspect, the vibration evaluation value z does not overlap the reference rotating machine in order to expand the application range of the rotating machine to be diagnosed by providing a certain range of variation as the vibration evaluation value z. The unit is set to several tens of units.

  As described above in detail, according to the present invention, it is possible to provide a method for diagnosing the soundness of a rotating machine by applying a sound signal obtained by sound measurement to a certain evaluation criterion.

It is a flowchart of the soundness diagnostic method of the rotary machine which concerns on this invention. 1 is a configuration diagram of a measuring device 1. FIG. It is a correlation diagram of a vibration evaluation standard and a sound evaluation standard.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  The present invention is in operation of a rotating machine having a rotating system such as a pump and a blower used in various facilities such as water and sewage facilities, river management facilities, waste treatment facilities, building facilities, septic tank facilities, power generation facilities, and other factories. This is a method for diagnosing the soundness of a rotating machine by measuring the generated sound. Specifically, as shown in the flowchart of FIG. 1, vibration and sound during operation of a reference rotating machine are measured in advance by the measuring device 1 shown in FIG. 2, and these vibration measurement values and sound measurement values are measured. Then, conversion coefficients a and b that have the largest correlation are obtained, and then the sound during operation of the rotating machine to be diagnosed is measured by the microphone 2 in the measuring device 1 shown in FIG. The sound value of the rotating machine is diagnosed by converting the value using the conversion coefficients a and b (sound evaluation value U) and applying the converted sound measurement value (sound evaluation value U) to the vibration evaluation standard. .

  In the measurement device 1, in the sound measurement, the signal measured by the microphone 2 is amplified to an appropriate amplitude by the input amplifier 3, then converted into a digital signal by the AD converter 4, and then taken into the personal computer P. . On the other hand, in the vibration measurement, the signal measured by the sensor 5 is amplified to an appropriate amplitude by the input amplifier 6 and then converted into a digital signal by the AD converter 7 and then taken into the personal computer P. Then, the signal processing device 8 performs various analyzes and diagnoses of the captured sound and vibration data, and records and stores them in the data recording device 9 such as the hard disk of the personal computer P, and the numerical values and charts of the analysis results are displayed. It is displayed on the monitor 10.

  As the standard rotating machine, it is possible to select a rotating machine in units of several tens of units, with the vibration evaluation values z of different models, models, sizes, etc. not overlapping, centering on the rotating machine that continuously performs diagnosis by sound signals. preferable. That is, it is preferable to select a reference rotating machine so that the vibration evaluation value z, which will be described in detail later, has a certain variation range, by measuring vibrations, thereby expanding the application range of the rotating machine to be diagnosed To be able to.

  Hereinafter, the diagnosis method will be described in more detail. The diagnosis method includes the following four steps.

(First step)
As a first step, vibration and sound during operation are measured in advance for a reference rotating machine, and the vibration measurement value is set as a vibration evaluation value z, and a residual analysis using an autoregressive model is performed from the sound measurement value. A sound residual power x and a correlation coefficient y in the regression map of the residual are obtained.

The autoregressive model is originally a technology used in predictive control systems, but in a system such as an inertial system that is expressed by a quadratic differential equation, if the motion from the past to the present is known, The motion can be accurately predicted. Specifically, this includes vertical movement of a weight suspended by a spring, movement of a pendulum, and the like.
Here, taking single string motion as an example, the equation for single string motion is expressed by the following equation (1).

When the above equation (1) is expressed in a difference format, the following equation (2) is obtained.

Therefore, it can be expressed by the following formula (3).

  That is, it can be seen that the displacement at time (i + 1) Δt is determined by the amplitude values at time iδt and time (i−1) Δt. Thus, the autoregressive model is a prediction model that can estimate the future from past and present information.

In the autoregressive model, as shown in the following equation (4), the current value (Y) is obtained by linear combination of the measured values (y _i-1 , y _i-2 , ..., y _iN ) of the past state quantities. _i ) is estimated. In the following formula (4), the subscript i means the time iδt when the measurement object is digitized at the discrete time interval δt.

In the above equation (4), coefficient sequences a ₁ , a ₂ , a ₃ ,..., _An are the regression coefficients of the autoregressive model, and the behavior of the system is determined by this number sequence.

The residual Q based on the autoregressive model is expressed by the following equation (5).

  Residual power x is used as an index representing the difference between the predicted value based on the autoregressive model and the actual value.

  Further, the residual at time i is plotted on the horizontal coordinate with the residual at time i as the horizontal axis and the residual at time i + 1 as the vertical axis, and regression mapping is performed. From this regression mapping, the residual at time i and the residual at time i + 1 are The correlation coefficient y in the regression map of the residual is used as an index representing the correlation of.

By the way, the sound generated from the device includes an air sound and a vibration sound, and basically the vibration sound is caused by the vibration of the side portion of the device in contact with the air. The general relationship between vibration and sound is W = ρcAv ² (W: sound output, ρ: density of air, c: speed of sound in air, A: area, v: vibration speed of surface) In addition, the sound pressure (sound output W) and the vibration speed (v) are in a proportional relationship. This indicates that the magnitude of vibration and the magnitude of sound are in a proportional relationship. However, this formula does not mention the frequency configuration. That is, conventionally, in the soundness diagnosis of equipment by vibration measurement, only the magnitude of vibration (overall value) is used as an evaluation scale, and the frequency configuration has not been discussed. On the other hand, in this diagnostic method, the frequency configuration during operation of the rotating machine is also evaluated by using the fact that the correlation coefficient y in the regression map of the residual changes when the frequency configuration changes. .

(Second step)
As a second step, the sound evaluation value u is set by linear weighting by multiplying the residual power x and the correlation coefficient y by the conversion coefficients a and b by the following equation (6).

Next, conversion coefficients a and b are determined under the condition that the square error Q between the vibration evaluation value z and the sound evaluation value u is minimum (correlation is maximum) by the following equation (7).

  As a method for obtaining the conversion coefficients (multiple regression coefficients) a and b of the above formula (7), it is preferable to use the least square method (multiple regression analysis method). Moreover, methods such as quantification type 1 can also be used.

(Third step)
After that, the sound during operation is measured for the rotating machine to be diagnosed, and the residual power X of the sound and the regression map of the residual using the residual analysis by the autoregressive model as described above from the sound measurement value. The correlation coefficient Y is obtained and converted into a sound evaluation value U by the following equation (8).

  The conversion coefficients a and b are values obtained from the above equation (7). As a result, a relationship with the reference rotating machine can be obtained.

(4th step)
Thereafter, the sound evaluation value U is applied to a vibration evaluation standard, and the soundness of the rotating machine is diagnosed according to the vibration evaluation standard.

  As the vibration evaluation standard, an original standard based on uniquely accumulated data or a standard proposed by various organizations can be used. In JIS B 0906 (ISO 10816-1), for each machine group, Since provisional standards for broadband vibration are established, it is preferable to use the evaluation standards defined therein.

  In the JIS standard, the vibration evaluation zone is classified into four zones according to the Rms value of the vibration speed, and is defined in the order of zone A, zone B, zone C and zone D from the sound side. Also in soundness diagnosis by sound measurement, it is classified into four zones (for example, level 0 to level 3), and a unique sound in which the range of the sound evaluation value U corresponding to the zone A to zone D of the JIS standard is set. It is also possible to create an evaluation standard and evaluate based on this. In creating such an evaluation standard, vibration evaluation is performed by plotting the results of simultaneous measurement of vibration and sound for a plurality of rotating machines on a correlation diagram between the vibration evaluation standard and the sound evaluation standard as shown in FIG. The range of the sound evaluation standard according to the standard can be determined.

  DESCRIPTION OF SYMBOLS 1 ... Measuring device, 2 ... Microphone, 3 ... Input amplifier, 4 ... AD converter, 5 ... Sensor, 6 ... Input amplifier, 7 ... AD converter, 8 ... Signal processing device, 9 ... Data recording device, 10 ... Monitor

Claims (4)

A method of diagnosing the soundness of a rotating machine by measuring sound,
Measure vibration and sound during operation for the reference rotating machine in advance, obtain a conversion coefficient that maximizes the correlation between these vibration measurement values and sound measurement values, and then operate for the rotating machine to be diagnosed The sound of the hour is measured, the sound measurement value is converted using the conversion coefficient, and the soundness of the rotating machine is diagnosed by applying the converted sound measurement value to the vibration evaluation standard. Diagnostic method for soundness of rotating machinery. The vibration and sound at the time of operation are measured for the reference rotating machine in advance, and the vibration measurement value is set as the vibration evaluation value z, and the residual power x of the sound is obtained from the sound measurement value using a residual analysis by an autoregressive model. And a first step of obtaining a correlation coefficient y in the regression map of the residual,
The sound evaluation value u is set by linear weighting by multiplying the residual power x and the correlation coefficient y by the conversion coefficients a and b by the following equation (6), and the vibration evaluation value by the following equation (7): a second step of determining the conversion coefficients a and b under the condition that the square error Q between z and the sound evaluation value u is minimized;
The sound during operation is measured for the rotating machine to be diagnosed, and the residual power X of the sound and the correlation coefficient Y in the regression map of the residual are obtained from this sound measurement value using the residual analysis by the autoregressive model. A third step of converting the sound evaluation value U into the following equation (8):
The soundness diagnostic method for a rotating machine according to claim 1, further comprising a fourth step of diagnosing the soundness of the rotating machine by applying the sound evaluation value U to a vibration evaluation standard.

  The soundness diagnosis method for a rotating machine according to any one of claims 1 and 2, wherein the vibration evaluation standard uses an evaluation standard defined in JIS B0906.   The soundness diagnosis method for a rotating machine according to any one of claims 1 to 3, wherein the reference rotating machine is set in units of several tens of units so that the vibration evaluation value z does not overlap.

Images