JP3598266B2 - Device abnormality diagnosis method and device - Google Patents

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a device abnormality diagnosis method and apparatus for identifying an abnormality position of an abnormal sound by using a device generated sound in the case of a device such as an automobile or an industrial machine.
[0002]
[Prior art]
Generally, the abnormality of the device is enlarged in order to continue the operation in a state where a small abnormality has occurred. Therefore, it is required to detect an abnormality of the device at an early stage and take an appropriate countermeasure.
[0003]
Conventionally, to detect abnormalities in equipment, measure the noise and vibration generated by the equipment, and compare the sound state of the equipment with its level and its frequency characteristics to detect The position is specified. This abnormality detection is effective when there is only one abnormal sound occurrence position, or when there is little sound other than the abnormal sound of the device, reflected sound, or the like.
[0004]
However, when there are a plurality of abnormal sound occurrence positions, or when sound (background noise) other than noise from the device reaches a preset evaluation point in the vicinity of the device, the position cannot be specified. Very difficult.
[0005]
In order to solve this problem, equipment-generated sound sensors are provided at a plurality of locations on the equipment, and the influence ratio (degree of contribution) is determined based on the equipment-generated sound (noise) energy ratio detected by these equipment-generated sound sensors. It has been proposed to identify the location of occurrence. This is described, for example, in JP-A-11-118592.
[0006]
The calculation of the influence ratio (degree of contribution) based on the energy ratio of the device-generated sound (noise) detected by the device-generated sound sensor is also described in JP-A-5-26722.
[0007]
[Problems to be solved by the invention]
In the related art, an abnormality occurrence position is specified by obtaining an influence ratio (contribution degree) based on a device generated sound (noise) energy ratio detected by a plurality of device generated sound sensors. For this reason, when there are a plurality of specific frequencies that are abnormal sounds, it is determined collectively and there is a problem that the generation position cannot be specified.
[0008]
An object of the present invention is to provide a device abnormality diagnosis method and apparatus capable of accurately specifying a position where an abnormal sound generated due to a device abnormality is generated.
[0009]
[Means for Solving the Problems]
A feature of the present invention is to perform a contribution analysis in which the frequency of an abnormal device sound at a predetermined evaluation point is specified by a sound level meter and an influence ratio (contribution degree) is determined for the abnormal device sound frequency specified by the sound level meter. It is to have done.
[0010]
According to the present invention, a contribution analysis is performed to determine an influence ratio (degree of contribution) for one or a plurality of abnormal sound frequencies of a device specified by a sound level meter installed at a predetermined evaluation point. The occurrence position of the abnormal sound can be accurately specified.
[0011]
BEST MODE FOR CARRYING OUT THE INVENTION
FIG. 1 shows an embodiment of the present invention.
[0012]
In FIG. 1, a machine tool 1 is provided with a plurality of device generated sound sensors 3a to 3d for taking in noise generated from the machine. FIG. 1 shows an example in which four sensors 3a to 3d are provided. The noise signals captured by these sensors 3a to 3d are input to the contribution analysis unit 5.
[0013]
In addition, background noise sensors 4a to 4d for taking in background noise other than the noise generated from the device 1 are installed around an evaluation point position previously set as a position for evaluating an abnormal sound, and these sensors 4a to 4d Is also input to the contribution analysis unit 5. At the evaluation point position, an evaluation point sensor 2 for taking in the arriving noise and outputting it as a target noise signal, and a sound level meter 7 for monitoring the time variation of the noise at the evaluation point position and performing frequency analysis are installed. I have.
[0014]
The target noise signal captured by the evaluation point sensor 2 is input to the contribution analysis unit 5. After a signal taken into the sound level meter 7 is subjected to frequency analysis, a specific frequency which causes an abnormal sound is extracted, and the extracted signal is applied to a personal computer (personal computer) 6 as a control calculation means.
[0015]
The information on the specific frequency of the abnormal sound is sent from the personal computer 6 to the contribution analysis unit 5, and is instructed to perform the calculation for the specific frequency band. The contribution analysis unit 5 performs convergence calculation using signals taken from the sensors 3a to 3d and 4a to 4d, and calculates the degree of influence of each sensor position noise on the evaluation point. The result of the calculation performed by the contribution analysis unit 5 is input to the personal computer 6 again. From this result, the location of the abnormal sound of the device is specified, and the influence of background noise is grasped.
[0017]
FIG. 2 shows an example of a detailed configuration diagram of the contribution analysis unit 5.
[0018]
In FIG. 2, one or a plurality of contribution analysis units 5 are provided near a bandpass filter 10 for a signal output from an evaluation point sensor 2 installed at an evaluation point, and a sound source (a device where abnormal sound is generated). The device-generated sound sensors 3a to 3d for detecting abnormal sounds, the background noise sensors 4a to 4d installed around the evaluation points, and the band-pass filters 11a to 11d and 12 to the output signals of the sensors 3a to 3d and 4a to 4d. 12d, delay circuits 13a to 13d and 14a to 14d for delaying the outputs of the band-pass filters 11a to 11d and 12a to 12d, and an A / D converter 15 for converting the outputs of the delay circuits 13a to 13d and 14a to 14d into digital values. , Adaptive filters 16a to 16d and 17a to 17d for adjusting the amplification, phase and the like of the output of A / D converter 15, adaptive filter 1 an addition circuit 18 for adding the outputs of a to 16d and 17a to 17d; a comparator 19 for comparing the output waveform of the addition circuit 18 with the output waveform of the bandpass filter 10; And a convergence calculating means 20 for executing a calculation for updating the coefficients of the adaptive filters 16a to 16d to 17a to 17d based on a predetermined adaptive algorithm so that the residual becomes equal to or smaller than a predetermined value.
[0019]
The adaptive algorithm for performing the convergence calculation of the contribution analysis unit 5 shown in FIG. 2 is well known to those skilled in the art, and for example, an algorithm described in Japanese Patent Application Laid-Open No. 5-26722 can be used.
[0020]
Now, a sound level meter 7 is installed at the evaluation point position, and a frequency analysis is performed to extract a specific frequency that is a cause of an abnormal sound. Information on the specific frequency is input to the contribution analysis unit 5 through the personal computer 6. In the contribution analysis unit 5, information of a specific frequency is sent to the band-pass filters 11a to 11d and 12a to 12d through which input signals from the sensors 3a to 3d and 4a to 4d pass. By specifying a frequency band, contribution analysis for a specific frequency band can be performed.
[0021]
In the contribution analysis, the coefficients of the adaptive filters 16a to 16d and 17a to 17d are automatically updated so that the output waveform of the band-pass filter 10 and the output waveform of the adding circuit 18 match, and the residual signal Is sufficiently small and reaches a predetermined value, the energy ratios of the output signals of the adaptive filters 16a to 16d and 17a to 17d at that time indicate the respective degrees of influence in the vicinity of the device-generated sound sensors 3a to 3d for abnormal sound detection, and The influence of other background noises can be determined.
[0022]
FIG. 3 is an operation flow chart up to identification of an abnormal sound occurrence position of the device according to the present invention.
[0023]
First, in step S1, the sound level meter 7 detects an abnormal sound. As shown in FIG. 4A, the sound level meter 7 determines from the measurement result that an abnormal sound has occurred at a certain time t1.
[0024]
Next, the sound level meter 7 extracts an abnormal sound frequency component in step S2, and performs frequency analysis on the abnormal sound detected in step S1. The sound level meter 7 performs a frequency analysis on the noise at the time t1 when the abnormal sound is detected in FIG. 4A, and detects the specific frequency f1 from the frequency analysis result as illustrated in FIG. 4B. 4A and 4B, the information of the time change of abnormal sound and the result of frequency analysis are input to the personal computer 6 and then input to the contribution analysis unit 5 as information for specifying a specific frequency. Is done.
[0025]
In step S3, the contribution analysis unit 5 starts a contribution analysis on the specific frequency extracted in step S2. The details of the analysis performed by the contribution analysis unit 5 are as shown in FIG. In step S4, it is determined whether or not the abnormal sound fluctuates from the time change of the sound level meter 7.
[0026]
If the abnormal sound is fluctuating, the process proceeds to step S5, and the contribution analysis unit 5 performs real-time analysis. The abnormal sound generation position is determined from the sensor position at which the degree of influence fluctuates following the fluctuation of the abnormal sound, as shown in step S6. The position of the occurrence of a fluctuating sound or an abnormal sound generated as a single sound is specified. As a result, the process proceeds to step S11, and the abnormal sound occurrence location identification of the device 1 is ended.
[0027]
Next, a case where the abnormal sound does not fluctuate will be described. In this case, the contribution analysis unit 5 calculates the degree of influence in step S8. As a result, as shown in FIGS. 5A and 5B, the degree of influence of each sensor is calculated and displayed on the personal computer 6.
[0028]
In FIG. 5A, since the degree of influence of the sensor 3d is large, it can be determined that an abnormal sound is generated near the device generated sound sensor 3d. Further, in FIG. 5B, since the influence of the sensors 4b and 4c is large, it can be determined that the abnormal sound is actually caused not by the device 1 but by other background noise.
[0029]
Such a determination can be made even when the factor causing the abnormal sound is not the device but the background noise.
[0030]
The occurrence position of the specific frequency range noise that causes the abnormal sound is identified by identifying the sensor having the greatest influence in step S9. Therefore, it is possible to determine the position of the abnormal sound occurrence from the device in step S10 which is continuous and unchanging from the display result of the personal computer 6, and in the final step S11, the abnormal sound occurrence position of the device 1 is specified.
[0031]
Note that this method is also effective when a plurality of specific frequencies represented by the specific frequency f1 in FIG. At this time, the contribution analysis is performed in a form where the band is narrowed down to each of the plurality of specific frequencies, and the sensor having a large influence is specified.
[0032]
FIG. 6 shows an analysis result of a conventional case in which contribution analysis is performed without performing frequency analysis on noise at the evaluation point position.
[0033]
Since there is no designation of a specific frequency for the band-pass filter of the contribution analysis unit 5, the contribution analysis is performed in a form including the entire frequency range. Although it is possible to extract a sensor position having a large degree of influence over the entire frequency range, it is not possible to determine which frequency range the sensor has a large effect on sound. Therefore, when there are a plurality of specific frequencies f1, f2,... As shown in FIG. 4B, it is difficult to specify the generation position for each specific frequency in the analysis including the entire frequency range.
[0034]
In this way, the position where the abnormality of the device occurs is specified. Since the specific frequency of the device abnormal sound is extracted in advance and the contribution analysis is performed, the specification of the position where the abnormal sound occurs is specified for each specific frequency band. This makes it possible to accurately specify the generation position.
[0035]
【The invention's effect】
According to the present invention, a contribution analysis is performed to determine an influence ratio (contribution degree) for one or more abnormal sound frequencies of a device specified by a sound level meter installed at a predetermined evaluation point. The occurrence position of the abnormal sound can be specified accurately.
[Brief description of the drawings]
FIG. 1 is a configuration diagram showing one embodiment of the present invention.
FIG. 2 is a detailed configuration diagram of an example of a contribution analysis unit.
FIG. 3 is a flowchart illustrating the operation of the present invention.
FIG. 4 is a characteristic diagram for explaining the present invention.
FIG. 5 is an example of a contribution analysis result of a specific frequency for explaining the present invention.
FIG. 6 is an example of a contribution analysis result according to a conventional technique.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Machine tool, 2 ... Evaluation point sensor, 3a-3d ... Equipment generated sound sensor, 4a-4d ... Background noise sensor, 5 ... Contribution analysis unit, 6 ... Personal computer, 7 ... Sound meter, 10, 11a-11d, 12a to 12d: band-pass filter, 13a to 13d, 14a to 14d: delay circuit, 15: A / D converter, 16a to 16d, 17a to 17d: adaptive filter, 18: adder, 19: comparator, 20 ... Convergence calculation unit.

Claims (3)

The background noise at a predetermined evaluation point is detected by a plurality of background noise sensors and the sound generated at a plurality of locations of the device is detected by a plurality of device generated sound sensors, and the output signals of the background noise sensor and the device generated sound sensor are output. After passing through the band-pass filter and passing through the adaptive filter, and is added. The residual signal between the added noise signal and the target noise signal output from the evaluation point sensor for detecting the noise at the evaluation point is a predetermined value. Determining the influence location by the output energy ratio of the background noise sensor and the device abnormal sound frequency component in the device abnormal sound frequency component when the background noise sensor and the device generated sound sensor converge, and specifying the generation position of the device abnormal sound. A sound level meter is installed at the position, the frequency of the abnormal sound of the device is specified by the noise signal detected by the sound level meter, and the band pass is determined based on the specified specific frequency. Apparatus abnormality diagnosis method being characterized in that to adjust the frequency band of the filter. A plurality of background noise sensors installed at predetermined evaluation points to detect background noise, a plurality of device generated sound sensors to detect generated sounds at a plurality of locations of the device, and a target noise signal by detecting noise at the evaluation points The evaluation point sensor to be output, the output signal of the background noise sensor and the output signal of the device generated sound sensor are fetched through a bandpass filter and added after passing through an adaptive filter, and the added noise signal and the target noise signal are added. When the residual signal converges to a predetermined value, the background noise sensor and the device-generated sound sensor determine the influence ratio based on the output energy ratio of the device abnormal sound frequency component to determine the position where the device abnormal sound is generated. Analyzing means, and a sound level meter installed at the evaluation point position, and specifying the frequency of the device abnormal sound based on the noise signal detected by the sound level meter; Apparatus abnormality diagnosis apparatus being characterized in that to adjust the frequency band of the band-pass filter based on a specific frequency. The background noise at a predetermined evaluation point is detected by a plurality of background noise sensors and the sound generated at a plurality of locations of the device is detected by a plurality of device generated sound sensors, and the output signals of the background noise sensor and the device generated sound sensor are output. After passing through the band-pass filter and passing through the adaptive filter, and is added. The residual signal between the added noise signal and the target noise signal output from the evaluation point sensor for detecting the noise at the evaluation point is a predetermined value. Determining the influence location by the output energy ratio of the background noise sensor and the device abnormal sound frequency component in the device abnormal sound frequency component when the background noise sensor and the device generated sound sensor converge, and specifying the generation position of the device abnormal sound. The frequency of the device abnormal sound is specified by the noise signal detected by the sensor, and the frequency band of the band-pass filter is adjusted based on the specified specific frequency. Apparatus abnormality diagnosis method being characterized in that so as to.

Images