JPH10181844A - Conveyor abnormality detection method and device therefor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To determine abnormality in a conveyor on the basis of abnormal sounds emitted therefrom of unspecific frequency bands with fluctuations in volume. SOLUTION: A conveyor abnormality detection method picks up sounds emitted from a conveyor at plural positions along it and filters out sounds in bands including mechanical sounds peculiar to the conveyor in the picked-up sounds to extract abnormal sounds from the picked-up sounds in S1 and S2, calculates the power spectra of the abnormal sounds to obtain values indicative of the intensities of the abnormal sounds from them in S3 as well as calculates the auto-regression coefficients of the abnormal sounds to obtain values indicative of the bandwidths of the abnormal sounds from them in S4, multiplies the intensity values and the bandwidth values of the abnormal sounds together to obtain abnormal sound evaluation values at the plural positions in S5, and detects a position where abnormality has occurred on the basis of the abnormal sound evaluation values in S6.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a conveyor abnormality detection method and apparatus for detecting conveyor abnormality from sound, and more particularly, to abnormality detection from abnormal sound in which the frequency band is unspecified and the sound volume fluctuates. The present invention relates to a conveyor abnormality detection method and apparatus.

[0002]

2. Description of the Related Art It has been conventionally performed to determine an abnormality of a line-shaped transporting machine (hereinafter, referred to as a conveyor) such as a belt conveyor based on sounds heard from a conveyor. The sound that can be heard from the conveyor has an abnormal sound related to some abnormality superimposed on the inherent mechanical sound that the conveyor originally emits. Although the mechanical sound peculiar to the conveyor is much larger than the mechanical sound peculiar to the conveyor, a skilled worker can distinguish the abnormal sound. However, depending on the type, scale and installation environment of the conveyor, it may be difficult or unfavorable for the patrol by the operator.

On the other hand, in a factory or the like, an acoustic sensor is conventionally mounted on a patrol robot or the like, and a spectrum of a sampled sound is analyzed. When a sound of a specific frequency exceeds a predetermined threshold value, it is determined that the sound is abnormal. Are known.

[0004]

However, there are various abnormalities in the conveyor, and when abnormal sounds caused by these abnormalities do not have a specific frequency, accurate abnormality determination may not be performed by the conventional method. Further, in a conveyor in which the mechanical sound changes in magnitude according to the driving situation, the abnormal sound also changes in magnitude accordingly, so that even if the abnormal sound is large to some extent, it may not always be a serious abnormality.

The gallery conveyor described below is for transferring a load such as coal, and has a belt formed in an endless belt shape for loading and circulating the load, and a predetermined interval in a longitudinal direction for supporting and sending the belt. And rollers provided. As an abnormality, a failure of a roller is common. Conventionally, an inspection is performed in which a worker patrols and hears abnormal noise of the rollers.

The characteristics of the sound of the gallery conveyor are as follows.
The inherent mechanical noise is mainly composed of the rotation noise of the roller, and its band is a low band of about 2 kHz. The abnormal sound occurs in a high frequency region having a higher frequency. The abnormal sound is a sound of squeaking of the bearings and a sound of the rollers resonating at the beginning, and a sound of the retainer breaking and the debris rubbing against the belts and rollers at the end. Since there is no need to replace the rollers in the initial stage, it is only necessary to be able to determine the end stage. However, the abnormal sound in the last stage is rarely biased to a specific frequency, and is characterized by a wide frequency band. The loudness of the inherent mechanical sound is much larger than that of the abnormal sound, and the loudness of the sound greatly differs between when the load is large and when the load is small.

[0007] Since this gallery conveyor transports coal and the like, the atmosphere in the line contains a lot of dust and is generally hot and humid. In addition, the gallery conveyor is installed at a steep inclination because of a difference in height between lines. The length of the gallery conveyor is as long as 200 to 800 meters. Therefore, the worker patrols the inclined scaffold over a long distance. In addition, the sound of the gallery conveyor is 120 dB or more, which is loud for humans to hear. Such a work environment is severe for workers, called 3K, and automation of abnormality determination is desired.

However, the conventional traveling robot determines that an abnormal sound is abnormal when a sound of a specific frequency exceeds a predetermined threshold, so that the abnormal sound does not become a specific frequency like the sound of a gallery conveyor. Further, it cannot be applied to a device having such a characteristic that the abnormal sound changes in magnitude depending on the load. Also, explosion-proof measures are required because there is a lot of dust in the atmosphere of the line, but explosion-proof measures are not considered for conventional patrol robots.

SUMMARY OF THE INVENTION It is an object of the present invention to solve the above problems and to provide a conveyor abnormality detection method and apparatus capable of judging abnormality from abnormal sound whose frequency band is unspecified and whose sound volume fluctuates. .

[0010]

SUMMARY OF THE INVENTION In order to achieve the above object, a method of the present invention collects sounds at a plurality of points along a conveyor, and removes a band containing a mechanical sound unique to the conveyor from the sounds. By extracting only the abnormal sound, the power spectrum of the abnormal sound is obtained, a value indicating the intensity of the abnormal sound is obtained from the power spectrum, and the autoregressive coefficient of the abnormal sound is obtained, and the abnormal value is obtained from the autoregressive coefficient. A value indicating the range of the sound range is obtained, and the product of the intensity value of the abnormal sound and the value of the range of the sound range is used as an abnormal sound evaluation value for each location. It is to detect.

A portion where the abnormal sound evaluation value exceeds a predetermined extraction threshold is extracted, and the extracted portions are weighted in descending order of the abnormal sound evaluation value, and the weighted evaluation value is determined in a predetermined manner. A location exceeding the threshold value may be determined as a failure occurrence location.

The above-mentioned sound sampling may be repeated a plurality of times, and the weighted evaluation value of each time may be averaged for each location, and this average value may be used for determining the location where an abnormality has occurred.

The apparatus further comprises a sound sampling means for sampling sounds at a plurality of locations along the conveyor, and a filter for extracting only abnormal sounds by removing a band including a mechanical sound unique to the conveyor from the sounds. And a sound intensity calculating means for obtaining a power spectrum of the abnormal sound to obtain a value indicating the intensity of the abnormal sound from the power spectrum, and obtaining an autoregressive coefficient of the abnormal sound to obtain an abnormal sound from the autoregressive coefficient. Means for calculating a range indicative of the value of the range of the range, and the product of the value of the intensity of the sound and the value of the range of the range as an abnormal sound evaluation value for each location; And an abnormality occurrence point detecting means for detecting an abnormality occurrence point.

[0014] The sound collecting means includes a moving vehicle disposed above the conveyor, a movement control means for reciprocating the moving vehicle along the conveyor, and a sound mounted on the moving vehicle and having directivity downward. A sensor and transmission means for transmitting a detection signal of the acoustic sensor to the filter may be provided.

[0015] The transmission means may be a radio transmitter for performing radio transmission.

[0016]

An embodiment of the present invention will be described below in detail with reference to the accompanying drawings.

FIG. 1 shows a processing procedure when the method of the present invention is applied to a gallery conveyor. In the gallery conveyor, the rotation sound of the rollers occupies most of the energy of the sound, and the band of the inherent mechanical sound is in a low band, and the abnormal sound is generated in a high band with a higher frequency. The purpose of this processing is to detect a portion where the roller abnormality has occurred from the abnormal sound. Therefore, sound is collected at a plurality of locations along the conveyor. By setting the location where sound is collected in the vicinity of each roller, it is possible to immediately identify a roller abnormal location if an abnormal occurrence location is detected from the sound capturing locations. In addition, one line of sound collection along the conveyor is regarded as one inspection, and this inspection is repeated.

First, the rotation noise of the roller is removed from the collected sound using a high-pass filter, and only the abnormal sound is extracted (S1). Sampling is performed at a predetermined sampling cycle in order to provide the abnormal sound to the roller abnormal portion detection processing (S2). Thereafter, the abnormal sound is processed as time-series data. When the time-series data for each sound sampling location is obtained in this way, the process enters the roller abnormal location detection process.

In the roller abnormal portion detecting process, first, the operation for obtaining the value of the intensity of the abnormal sound (S3) and the operation for obtaining the value of the width of the sound range of the abnormal sound (S4) are independently performed. Next, an operation (S5) for obtaining an abnormal sound evaluation value for each location by multiplying the value of the sound intensity and the value of the breadth of the sound range is performed. Is performed (S6). After the end of the detection processing, the result is displayed (S7).

The calculation processing of the intensity of the abnormal sound will be described with reference to FIG.

In the calculation of the intensity of the abnormal sound, a time waveform is cut out by extracting the number of samples necessary for the spectrum calculation, here 1024, from the time series data of the sound sampling location (S31). This cutout is performed a plurality of times for one location, here, five times. A power spectrum is calculated for the time waveform composed of 1024 samples (S32).

[0022]

(Equation 1)

Here, Ck is the power spectrum (complex number) of the k-th divided band, and N is the number of divided bands.

The above operation can be performed using the FFT algorithm. The power spectrum thus obtained is obtained by dividing the abnormal sound frequency band into N parts and expressing the energy components of each divided band. Next, an overall value is calculated by summing the absolute values of the N frequency components (S33).

[0025]

(Equation 2)

Here, Oa is an overall value.

The overall value thus obtained is
This is a value indicating the magnitude of the energy of the abnormal sound. The overall value is converted into a unit of sound intensity, and a sound intensity value having a dimension of W / m ² is used (S34).

[0028]

(Equation 3)

Here, Vcal is a microphone calibration voltage, ρ is density, c is sound speed, P is sound pressure, I
Is the intensity of the sound.

The value of the sound intensity at the sound sampling location is averaged for the number of cutouts up to this time (S35).
The extraction is continued until the averaging process is completed for all the number of extractions for the sound sampling location. When the number of cutouts reaches the total number of times, for example, five, the same is repeated for the next sound sampling point. When the process of obtaining the value of the sound intensity and averaging the same for the number of cutouts is completed for the entire sound collection location, the calculation process of the intensity of the abnormal sound is completed.

The processing for calculating the width of the sound range will be described with reference to FIG.

In the calculation of the range of the abnormal sound range, the number of samples required for the autoregressive calculation, 1024 in this case, is extracted from the time series data of the sound sampling location.
A time waveform is cut out (S41). This cutout is performed a plurality of times for one location, here, five times.
An auto-regression coefficient is calculated for the time waveform including 1024 samples (S42).

Here, the auto-regression coefficient is a coefficient of an auto-regression equation. The auto-regression equation is such that the current value sample of the time series data is represented by a linear equation of m past samples.

[0034]

(Equation 4)

[0035] Here, a ₁ ~a _m is autoregressive coefficients, x _i
Is discrete data forming an autoregressive equation, and the time-series data x (t) can be expressed as {x _i (t = iΔt)}. Also, n (t) is a random input that becomes white noise.

[0036] When the self-regression coefficients from the time-series data a ₁ ~a _m
Can be performed using the Brug algorithm. Here, it is assumed that m = 32, and a 32nd-order autoregressive coefficient is obtained.

Next, the squares of the m-th order autoregressive coefficient are summed (S43).

[0038]

(Equation 5)

Here, S is the sum of squares of the autoregressive coefficient.

The sum of squares S of the autoregressive coefficient is a value indicating the range of the abnormal sound range. That is, if the frequency component of the corresponding time waveform is biased in frequency, the specific autoregressive coefficient is large, but the other autoregressive coefficients are small, so that the sum of squares S is small. On the other hand, if the frequency components are widely present, the plurality of auto-regression coefficients increase, so that the sum of squares S increases.

The sum of squares of the auto-regression coefficient (the range of the abnormal sound range) at the sound sampling location is averaged for the number of cutouts up to this time (S44). The extraction is continued until the averaging process is completed for all the number of extractions for the sound sampling location. The number of cutouts is all times, for example, 5
When the number of times has been reached, the same is repeated for the next sound sampling point. When the process of obtaining the range of the abnormal sound range and averaging the same for the number of cutouts is completed for the entire sound collection location, the calculation of the range of the abnormal sound range is completed. However, here, the values of the widths of the sound ranges at all points are normalized so that the maximum value becomes 1 (S45).

With reference to FIG. 4, a description will be given of a calculation for detecting a location where an abnormality has occurred based on an abnormal sound evaluation value.

First, in S5 of FIG. 1, an abnormal sound evaluation value is obtained by multiplying the value of the sound intensity by the value of the breadth of the sound range. That is, the average value of the sound intensity obtained in the processing of FIG. 2 is multiplied by the average sum of squares of the autoregressive coefficients obtained in the processing of FIG.

The abnormal sound evaluation value of each sound sampling position is sequentially extracted, and if there is a position where the abnormal sound evaluation value exceeds a predetermined extraction threshold, the position is stored and the number is counted (S51). . When it finishes up to the whole sound collection point,
This means that all the sound sampling locations where the abnormal sound evaluation value exceeds the extraction threshold have been extracted.

The extracted parts are sorted in descending order of the abnormal sound evaluation value (S52). Then, weighting is performed in the order in which they are arranged by the sorting process (S53). The weighted value is called a weighted evaluation value. Here, the present invention is configured to repeatedly perform this inspection with one line of sound collection along the conveyor as one inspection, and when the weighted evaluation value of this time is determined, The evaluation value and the current weighted evaluation value are averaged, and this average value is defined as the degree of abnormality. Accordingly, the degree of abnormality is updated each time, and becomes an average of a plurality of trends.

In determining the location where an abnormality has occurred, a location that exceeds a predetermined determination threshold is determined as the location where an abnormality has occurred. If this determination is YES, a roller abnormality location can be identified corresponding to the sound collection location (S54). At this time, an alarm is immediately generated (S55).

Next, FIG. 5 shows the data flow according to the processing procedures of FIGS. 1 to 4 and intermediate data obtained by calculation, and the present invention will be described with reference to FIG.

As shown in FIG. 5 (a), the raw data obtained from the sound sampling means (described later) at the sound sampling point is passed through a high-pass filter 51 having a cut-off frequency of 2.5 kHz to reduce the roller rotation sound. Removed. The belt speed of the gallery conveyor is 220 m / min, and the roller rotation cycle of each roller is 140 msec. The A / D converter 52 continuously collects 5120 samples at a sampling frequency capable of obtaining waveform information up to 10 kHz. This time-series data is provided to the sound intensity calculating means 53 written as FFT and the overall value and to the range calculating means 54 written as the sum of squares of the autoregressive coefficient. In the sound intensity calculation means 53, 102
Four samples are taken out five times, and each time, a power spectrum calculation, an overall value calculation, a unit conversion into a sound intensity are performed, and finally the five times are averaged. According to the range calculating means, it is necessary to take out 1024 samples.
The calculation is performed twice, and the autoregression coefficient calculation and the sum of squares calculation of the 32nd order are performed each time, and finally the five times are averaged.

As shown in FIG. 5B, a graph 501 in which the horizontal axis represents the sound sampling point n and the vertical axis represents the sound intensity, the sound intensity of 2.5 kHz or more at each sound sampling point. You can see the difference. On the other hand, in the graph 502 in which the sound sampling point n is plotted on the horizontal axis and the sum of squares of the autoregression coefficient is plotted on the vertical axis, the difference in the range of the abnormal sound range at each sound sampling point (difference in time waveform).
I understand. In the graph 503 in which the sum of squares has been normalized by the normalizing means 54, the difference in the range of the sound range can be seen as a ratio. The multiplication (integration) of the sound intensity value and the normalized range value by the abnormal sound evaluation value calculating means 55 corresponds to weighting the range of the sound range to the sound intensity. .
In the graph 504 in which the sound sampling point n is plotted on the horizontal axis and the obtained abnormal sound evaluation value is plotted on the vertical axis, the difference between the abnormal sound evaluation values at each sound sampling point can be seen. This abnormal sound evaluation value can be said to be an index representing the sound intensity and waveform characteristics.

As shown in FIG. 5 (c), a graph 505 is plotted on the abscissa of the sound sampling location n and on the ordinate of the abnormal sound evaluation value.
When the extraction threshold value 56 is written by a broken line, it is understood that a portion having a large abnormal sound evaluation value is extracted. In this figure, the sorting process is omitted, and therefore the ranking is added instead. That is, they are numbered 2, 1, and 3 in order from the leftmost sound collection point. Here, the weighting means 57, for example, 100 at the position where the abnormal sound evaluation value is the maximum, 50 at the second position,
A weighted evaluation value (unit:%) of 20 for the third location and 0 for the fourth and subsequent locations is provided. Therefore,
In a graph 506 in which the sound sampling point n is plotted on the horizontal axis and the weighted evaluation value is plotted on the vertical axis, the second, first, and third sound sampling points have values of 50, 100, and 20, respectively. The sound sampling locations are all zero. The average of the weighted evaluation values for each sound sampling location for sound sampling in a plurality of tests is the degree of abnormality, and the graph 507 in which the horizontal axis indicates the sound sampling point n and the vertical axis indicates the degree of abnormality, It can be seen that, when sound collection is repeatedly performed, a sound collection part having a higher rank of the abnormal sound evaluation value always shows a large abnormal degree. Therefore, the judgment threshold value 58
Is set to 50% in this case, and a sound sampling location indicating an abnormality degree of 50% or more is determined as an abnormal occurrence location.

According to the conveyor abnormality detecting method of the present invention,
Without using the power spectrum of a specific frequency for abnormality determination,
Since the strength of the abnormal sound was determined from the overall value of the power spectrum, it can be applied to the detection target whose frequency of the abnormal sound is unspecified. In addition, abnormal sounds having a wide frequency band can be mainly inspected. In addition, since the judgment is not made immediately from the magnitude of the abnormal sound evaluation value, weighting based on the rank is performed and the judgment is made based on the weighted evaluation value. Thus, only the portion where the abnormal sound is remarkable can be determined.

Next, the conveyor abnormality detecting device will be described.

As shown in FIG. 6, a gallery conveyor to be detected includes a belt 61 formed in an endless belt shape for loading and circulating a load, and a predetermined lengthwise interval for supporting and sending out the belt 61. In this embodiment, rollers 62 are provided at a pitch of 1 meter. The roller 62 includes a carrier roller 62a for an upper-stage forward path and a return roller 62b for a lower-stage backward path.

The conveyor abnormality detecting device extracts sound at a plurality of points along the conveyor, and extracts only abnormal sounds by removing a band including a mechanical sound unique to the conveyor from the sounds. A filter, a sound intensity calculating means for obtaining a power spectrum of the abnormal sound to obtain a value indicating the intensity of the abnormal sound from the power spectrum, and an autoregressive coefficient of the abnormal sound for obtaining the autoregressive coefficient of the abnormal sound. Means for calculating a range indicative of the width of the range of the sound, and an abnormal sound evaluation value which sets the product of the value of the intensity of the sound and the value of the range of the sound to an abnormal sound evaluation value for each location Calculating means; extracting means for extracting a portion where the abnormal sound evaluation value exceeds a predetermined extraction threshold; weighting means for weighting the extracted portions in descending order of the abnormal sound evaluation value; value And abnormality determining means for determining a place beyond the determined threshold defined because an abnormality occurrence location, and alarm generating means for generating an alarm by the determination result, and a display means for displaying a processing result or course. Here, the sound intensity calculating means, the sound area width calculating means, the abnormal sound evaluation value calculating means, the extracting means, the weighting means, the abnormal determining means, the alarm generating means and the displaying means are stored as software in a computer 64 such as a personal computer. Have been. The display is performed using the monitor of the computer 64. The computer 64 can notify the operation room of the gallery conveyor of the detection result.

The sound collecting means 63 includes a movable carriage 65 disposed above the conveyor, a movement control means 66 for reciprocating the movable carriage 65 along the conveyor, and a directional microphone (downwardly mounted on the movable carriage). Acoustic sensor)
67, and a cable (transmission means) 69 for transmitting the detection signal of the microphone 67 to the acoustic amplifier 68.
The filter is housed in the acoustic amplifier 68. The movement control means 66 includes a guide rail 651 provided in parallel with the conveyor to guide the movable cart 65, a wire (not shown) circulated along the guide rail 651, and a pulley 652 for reciprocating the wire. And a controller 653 for controlling the rotation of the pulley 652. The moving carriage 65 is fixed to a wire and moves by the rotation of the pulley 652.

The detection process executed by the conveyor abnormality detection device is as described above.
The sound collecting means 63 will be mainly described. First, the moving carriage 65 moves from one end to the other end of the moving range without stopping halfway at a speed of 0.1 m / s. During this time, the data of the sound picked up by the microphone 67 is transmitted to the acoustic amplifier 68 via the cable 69, and is sampled after preprocessing by the filter in the acoustic amplifier 68.

Assuming that the time required for the roller 62 of the gallery conveyor to make one rotation is 150 msec, at least 15 seconds are required in order to avoid the noise generated by the defective portion of the roller 62.
Since it is necessary to provide a waveform of 0 msec to the roller abnormal portion detection processing, the number of samples cut out from the time series data is determined according to the sampling cycle. On the other hand, the speed of the mobile trolley 65 is sufficiently low,
Sound collection at almost the same position is possible without stopping halfway.

In the conveyor abnormality detecting apparatus according to the present invention, since the movable carriage 65 is disposed above the conveyor and the directional microphone 67 is attached to the movable carriage 65 downward, external noise, especially adjacent rollers The sound from 62 can be cut off. That is, the sound of one roller 62 can be reliably inspected. Further, the carrier roller 62
a and the return roller 62b are arranged at substantially the same upper and lower positions.
The sounds from the a and 62b can be simultaneously sampled and an abnormality can be detected collectively.

Further, since no power unit is mounted on the movable carriage 65, it is excellent in explosion-proof measures in a dust environment.

Next, another embodiment of the sound sampling means will be described.

As shown in FIG. 7, the mobile trolley 65 is equipped with a wireless transmitter 71 for wirelessly transmitting a detection signal of the microphone 67. The wireless transmitter 71 includes a transmitter 72 and a transmission antenna 73. The transmitting antenna 73 is attached so as to be close to and opposed to a long receiving antenna 74 laid along the guide rail 651.

As shown in FIG. 8, a receiver 81 is connected to one end of a receiving antenna 74, and a computer 64 is connected to the receiving antenna 74 via an interface 82.
The data can be taken into.

In this embodiment, a cable for transmitting the detection signal of the microphone 67 is not required. When a cable 69 is connected to the movable carriage 65 as shown in FIG. 7, a cable handling mechanism is necessary. Therefore, in the embodiment of FIG. 8, not only the cable but also the accompanying cable handling mechanism is omitted. Since the entire structure is reduced in weight by omitting, the struts and the like can be simplified. Further, since there is no cable, there is no restriction on the moving distance, and the present invention can be easily applied to a long conveyor.

Next, the process after the detection of the roller abnormal part will be described.

When an abnormal roller portion is specified in S54, an alarm is generated in S55 immediately. After the end of the detection processing, the result is displayed in S7. As the display contents, the results shown in FIG. 5 and the graph of the intermediate data are displayed. Further, the roller design is displayed in red on the horizontal axis of the graph, that is, at a position corresponding to the sound sampling location. Further, in the conveyor abnormality detection device according to the present invention, the movable carriage 65 moves to the abnormal roller and stops. Thereby, the worker can be guided to the site. The operator only needs to closely inspect the roller where the moving carriage 65 is stopped.

[0066]

The present invention exhibits the following excellent effects.

(1) Workers are released from a harsh environment by automating the patrol of a long and long conveyor line due to the presence of dust at high temperature and high humidity.

(2) It is possible to automate the discrimination of sounds requiring skill, and to quantitatively diagnose abnormalities.

(3) Since the power spectrum and the auto-regression coefficient are used, the amount of data to be processed may be smaller as compared with the time waveform pattern recognition or the like.

(4) Since the value of the intensity of the abnormal sound and the value of the range of the sound range are evaluated in combination, the influence of narrow-band noise (such as paging) is reduced, and the detection accuracy is improved. .

(5) Since the weighting is performed in descending order of the abnormal sound evaluation value, a portion where the abnormal sound is remarkable can be extracted even when the sound volume is large as a whole.

[Brief description of the drawings]

FIG. 1 is a flowchart showing a processing procedure of a method of the present invention.

FIG. 2 is a flowchart showing a procedure for calculating the intensity of an abnormal sound according to the present invention.

FIG. 3 is a flowchart showing a procedure for calculating a range of a sound range according to the present invention.

FIG. 4 is a flowchart showing a processing procedure for detecting an abnormal occurrence location based on an abnormal sound evaluation value according to the present invention.

FIG. 5 is a data flow diagram showing a data flow according to the processing procedure of the present invention and intermediate data obtained by calculation.

FIG. 6 is a configuration diagram of a conveyor abnormality detection device according to the present invention.

FIG. 7 is a partial configuration diagram of the conveyor abnormality detection device of the present invention.

FIG. 8 is a configuration diagram of a conveyor abnormality detection device of the present invention.

[Explanation of symbols]

 Reference Signs List 61 Belt 62 Roller 63 Sound sampling means 65 Moving carriage 66 Movement control means 67 Microphone (acoustic sensor) S3 Calculation for finding intensity of abnormal sound S4 Calculation for finding value of sound range of abnormal sound S5 Abnormal sound evaluation value S6 Calculation for detecting the location where an error has occurred

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Shigeki Murayama 3-1-1-15 Toyosu, Koto-ku, Tokyo Ishikawajima-Harima Heavy Industries Co., Ltd. Tojin Technical Center (72) Inventor Tsutomu Hoshii 1-chome Mori, Koto-ku, Tokyo 19-10 Ishikawajima Harima Heavy Industries, Ltd.Koto Office

Claims (6)

[Claims] 1. A method according to claim 1, wherein sound is collected at a plurality of locations along the conveyor, and a band including a mechanical sound unique to the conveyor is removed from the sound to extract only an abnormal sound, and a power spectrum of the abnormal sound is obtained. A value indicating the intensity of the abnormal sound is obtained from the power spectrum, and a value indicating the range of the sound range of the abnormal sound is obtained from the autoregression coefficient of the abnormal sound. A conveyor abnormality detection method, characterized in that a product of a soundness value and a value of a range of a sound range is used as an abnormal sound evaluation value for each location, and an abnormal occurrence location is detected based on the abnormal sound evaluation value. 2. A portion where the abnormal sound evaluation value exceeds a predetermined extraction threshold is extracted, and the extracted portions are weighted in descending order of the abnormal sound evaluation value, and the weighted evaluation value is determined in advance. 2. The conveyor abnormality detection method according to claim 1, wherein a location exceeding the determination threshold is determined as an abnormality occurrence location. 3. The conveyor abnormality according to claim 2, wherein the sound sampling is repeated a plurality of times, and the weighted evaluation value of each time is averaged for each location, and the average value is used for determining an abnormality occurrence location. Detection method. 4. A sound sampling means for sampling sounds at a plurality of points along a conveyor, a filter for extracting only abnormal sounds by removing a band including a mechanical sound unique to the conveyor from the sounds, and a filter for extracting abnormal sounds. A sound intensity calculating means for obtaining a power spectrum of the sound to obtain a value indicating the intensity of the abnormal sound from the power spectrum; and a self-regression coefficient for the abnormal sound and obtaining a wider sound range of the abnormal sound from the self-regression coefficient. Means for calculating the width of the range for obtaining a value indicating the loudness, and the product of the value of the intensity of the sound and the value of the width of the range as an abnormal sound evaluation value for each location. A conveyor abnormality detection device, comprising: an abnormality occurrence point detection means for detecting an abnormality. 5. A sound pickup means comprising: a movable carriage disposed above a conveyor; a movement control means for reciprocating the movable carriage along the conveyor; and a directivity downwardly mounted on the movable carriage. The conveyor abnormality detecting device according to claim 4, further comprising: an acoustic sensor having the sensor; and a transmission unit configured to transmit a detection signal of the acoustic sensor to the filter. 6. The conveyor abnormality detection device according to claim 5, wherein the transmission unit is a wireless transmitter that performs wireless transmission.