JP7563582B2 - Abnormal part detection device, abnormal part detection method, and abnormal part detection system - Google Patents

Description

The present invention relates to an abnormal part detection device, an abnormal part detection method, and an abnormal part detection system.

There is a known invention for detecting sound and locating the position of the sound source (Patent Document 1). The invention described in Patent Document 1 converts information about the sound source, including the position of the sound source, into visible information and displays it on a display.

JP 2004-77277 A

A vehicle is densely packed with numerous parts. When an abnormal sound is emitted from a part, it is required to pinpoint the position of the part with high accuracy, but there is no mention of this in Patent Document 1.

The present invention has been made in consideration of the above-mentioned problems, and an object of the present invention is to provide an abnormal part detection device, an abnormal part detection method, and an abnormal part detection system that are capable of accurately identifying the position of a part that is emitting an abnormal sound.

An abnormal part detection device according to one embodiment of the present invention acquires sound data emitted by a vehicle, identifies the part generating the sound based on the acquired sound data, determines whether the sound data is abnormal based on sound characteristics corresponding to the identified part, and if it is determined that the sound data is abnormal, outputs information indicating that an abnormality has occurred.

According to the present invention, it is possible to accurately identify the location of a part that is emitting an abnormal sound.

FIG. 1 is a schematic diagram of an abnormal part detection system according to a first embodiment of the present invention. FIG. 2 shows an example of the configuration of an abnormal part detection device according to a first embodiment of the present invention. FIG. 3 is a diagram for explaining an example of characteristics of sound data caused by the shape, dimensions, material, etc. of an object. FIG. 4 is an image showing the positions of the parts. FIG. 5 is a flowchart illustrating an example of the operation of the abnormal part detection device according to the first embodiment of the present invention. FIG. 6 shows an example of the configuration of an abnormal part detection device according to the second embodiment of the present invention. FIG. 7 is an image showing the positions of the parts. FIG. 8 is a flowchart illustrating an example of the operation of the abnormal part detection device according to the second embodiment of the present invention.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings. In the description of the drawings, the same parts are given the same reference numerals and the description will be omitted.

First Embodiment
An example of the configuration of an abnormal part detection system according to a first embodiment will be described with reference to Figures 1 and 2. As shown in Figure 1, the abnormal part detection system is composed of a vehicle 10, a microphone 20 that acquires sounds emitted from the vehicle 10, and a computer 30. The sounds emitted from the vehicle 10 are converted by the microphone 20 into electrical signals and treated as "sound data." As shown in Figure 1, the sound data is output from the microphone 20 to the computer 30.

The details of the computer 30 will be described with reference to FIG. 2. The computer 30 is a terminal that processes sound data acquired from the microphone 20. The computer 30 includes, as an example, a CPU (controller 40), a memory (not shown), a storage device 50, and a display 60. The controller 40 loads various programs stored in the storage device 50 into the memory and executes various commands included in the programs. The storage device 50 is a storage medium such as an SSD (Solid State Drive) or an HDD (Hard Disk Drive). The controller 40 includes, as a plurality of information processing functions, a sound data acquisition unit 41, a sound source position identification unit 42, a sound characteristic acquisition unit 43, a part identification unit 44, an abnormality determination unit 45, an image generation unit 46, and an image output unit 47. The location where the computer 30 is installed is not particularly limited, but as an example, the computer 30 is installed in a repair shop.

The sound data acquisition unit 41 acquires sound data from the microphone 20. The sounds emitted from the vehicle 10 include sounds during inspection, sounds during test driving, sounds during normal driving, etc. The inspections include horn inspection, brake inspection, lamp inspection, etc.

The sound source position identifying unit 42 identifies the sound source position using the sound data acquired from the microphone 20. If there are multiple microphones 20, the sound source position can be identified by a method called a microphone array. One example of the method is a method of identifying the sound source position from the difference in arrival time of multiple sound data. This method is not limited to this method, and various algorithms are well known. In addition, the sound source position can be identified even if there is only one microphone 20. Therefore, the number of microphones 20 may be one or more. The identified sound source position is given as three-dimensional coordinates. Here, the three-dimensional coordinates of the identified sound source are (X0, Y0, Z0).

The sound characteristic acquisition unit 43 analyzes the sound data acquired from the microphone 20 to acquire the characteristics of the sound data. The characteristics of the sound data include sound pressure (dB) and frequency (Hz). An example of an analysis method is FFT (Fast Fourier Transform). The frequency includes a fundamental frequency and a harmonic corresponding to the fundamental frequency. The fundamental frequency means the frequency of the lowest frequency component when the sound data converted into an electric signal is expressed as a composite of sine waves. However, the definition of the fundamental frequency is not limited to this. The fundamental frequency is sometimes called a fundamental wave. A harmonic is a higher-order frequency component that is an integer multiple of the fundamental frequency. A harmonic is sometimes called an overtone. The characteristics of the sound data differ depending on the shape, dimensions, material, etc. of the object. An example of the characteristics of the sound data due to the shape, dimensions, material, etc. of the object is shown in FIG. 3. The shapes of the object include a rod shape, a plate shape, etc. Even if the shapes are similar, the sounds emitted by the parts differ depending on the dimensions and material (iron, aluminum, etc.).

The part identification unit 44 identifies the part that is emitting the sound. An example of a part identification method will be described. First, the part identification unit 44 acquires the characteristics of the sound data analyzed by the sound characteristic acquisition unit 43. The acquired characteristic of the sound data is called "characteristic A." Next, the part identification unit 44 refers to the storage device 50 and searches for a part having a characteristic of sound data similar to "characteristic A."

Here, the data stored in the storage device 50 will be described. In the storage device 50, the parts mounted on the vehicle are linked to the characteristics of the sound data of the parts, and stored. In other words, the characteristics of the sound data are linked to each part. Such linking is performed in advance. Each part is also linked to information indicating the position where it is placed. In other words, the characteristics of the sound data and position information are linked to each part, and the linked data is stored in the storage device 50. The characteristics of the sound data also include a fundamental frequency based on the shape, dimensions, and material of the part, and harmonic characteristics of the fundamental frequency.

Assume that the part identification unit 44 references the storage device 50 and detects two characteristics similar to "characteristic A." The two characteristics similar to "characteristic A" are called "characteristic A1" and "characteristic A2." The part linked to "characteristic A1" is called "part A1," and the part linked to "characteristic A2" is called "part A2." The position information linked to "part A1" is called "X1, Y1, Z1." The position information linked to "part A2" is called "X2, Y2, Z2."

The part identification unit 44 compares "characteristic A" with "characteristic A1". The part identification unit 44 also compares "characteristic A" with "characteristic A2". The comparison is made on position information. The root mean square error is used to compare the position information. If the error between "characteristic A" and "characteristic A1" is P1, the error P1 is expressed by Equation 1.

If the degree of agreement between “characteristic A” and “characteristic A1” is P2, the degree of agreement P2 is expressed by Equation 2.

Similarly, if the error between “characteristic A” and “characteristic A2” is Q1, the error Q1 is expressed by equation 3.

If the degree of agreement between “characteristic A” and “characteristic A2” is Q2, the degree of agreement Q2 is expressed by Equation 4.

The part identification unit 44 identifies the part with the higher degree of agreement, P2 or Q2, as the part that emitted the sound. If P2>Q2, the part that emitted the sound is identified as "part A1". If P2<Q2, the part that emitted the sound is identified as "part A2". Note that if three or more characteristics similar to "characteristic A" are searched for instead of two, the part with the highest degree of agreement is identified as the part that emitted the sound. As another identification method, since it is possible to identify the sound source position using sound data acquired from the microphone 20 as described above, the part identification unit 44 may identify the part that emitted the sound using sound data acquired from the microphone 20.

The abnormality determination unit 45 determines whether the sound data of the part identified by the part identification unit 44 is abnormal or not. "Abnormal" means not normal. "Abnormal" is a concept that includes not only failures that prevent a part from performing its original function, but also problems that do not result in failure and signs of failure. An example of an abnormality determination method will be described. Here, it is assumed that P2>Q2, and the part that emitted the sound is identified as "part A1". The characteristic of the sound data acquired from the microphone 20 is "characteristic A", and the characteristic of "part A1" stored in the storage device 50 is "characteristic A1". This "characteristic A1" indicates the characteristic of the sound data when "part A1" is normal.

The abnormality determination unit 45 compares "characteristic A" with "characteristic A1" and calculates the similarity as a numerical value. If the calculated numerical value is equal to or less than the threshold value (if the similarity is small), the sound data of "part A1" is determined to be abnormal. On the other hand, if the calculated similarity is greater than the threshold value (if the similarity is large), the sound data of "part A1" is determined to be normal. Cluster analysis, for example, is used to calculate the similarity. Note that the method of abnormality determination is not limited to using the similarity. There is also a method of evaluating the degree to which two pieces of data are similar by the closeness of the distance.

In this way, the abnormality determination unit 45 uses the characteristics of the sound data to determine whether the sound data of the part is abnormal or not. The abnormality determination unit 45 outputs a signal indicating that the sound data of "part A" is abnormal to the image generation unit 46.

When the abnormality determination unit 45 determines that the sound data of "part A" is abnormal, the image generation unit 46 generates an image showing the position of "part A". An example of an image generated by the image generation unit 46 is shown in FIG. 4. Reference numeral 80 shown in FIG. 4 is an image generated by the image generation unit 46. Hereinafter, it will be referred to as image 80. In FIG. 4, "part A" indicates an alternator. Reference numeral 70 is an image simulating a vehicle body. Reference numeral 71 is an image simulating an arrangement structure of a large number of parts mounted on a vehicle. When the arrangement structure is designed by three-dimensional CAD, reference numeral 71 may be a CAD drawing. Hereinafter, reference numeral 71 will be described as a CAD drawing. As shown in FIG. 4, an image 72 showing an alternator is extracted from the CAD drawing 71 and enlarged. As shown in FIG. 4, the image 70 simulating a vehicle body is subjected to a transparent process, so that a user (e.g., a person who performs repairs) can visually check the CAD drawing 71. Furthermore, since an image 72 showing the alternator is extracted from the CAD drawing 71 and enlarged, the user can grasp at a glance the part (here, the alternator) where the abnormality is occurring and the position of that part. The image generating unit 46 may add sound pressure information and frequency information to the image 72 showing the alternator. In the example shown in Fig. 4, the sound pressure is 86 (dB) and the fundamental frequency is 150 (Hz). As shown in Fig. 7, which will be described later, the image generating unit 46 may also display the presence or absence of overtones in Fig. 4.

The storage device 50 stores an image 70 simulating a vehicle body, a CAD drawing 71, and an image 72 showing an alternator. The image generation unit 46 generates an image 80 shown in FIG. 4 by referring to the storage device 50. Of course, the storage device 50 also stores images showing other parts in addition to the alternator. The image generation unit 46 outputs the generated image 80 to the image output unit 47.

Although it has been explained that the image 70 simulating a vehicle body has been subjected to a transparency process, it is not necessary that the entire image 70 has been subjected to a transparency process. For example, the transparency process does not need to be applied to the parts other than the CAD drawing 71. The information necessary for the user is the part where the abnormality has occurred and its location. If these two pieces of information can be grasped, other information can be said to be unnecessary information. Therefore, the transparency process may be applied only to the part in the CAD drawing 71. This prevents the user from being provided with unnecessary information.

The image output unit 47 outputs the image 80 generated by the image generation unit 46 to the display 60. By looking at the image 80 displayed on the display 60, the user can grasp at a glance the part in which an abnormality has occurred and the position of that part.

Next, an example of the operation of the abnormal part detection device (computer 30) according to the first embodiment will be described with reference to the flowchart of FIG.

In step S101, the sound data acquisition unit 41 acquires sound data from the microphone 20. The sound data acquisition unit 41 outputs the acquired sound data to the sound source position identification unit 42 and the sound characteristic acquisition unit 43. The process proceeds to step S103, where the sound source position identification unit 42 identifies the sound source position using the sound data acquired from the sound data acquisition unit 41. The sound source position identification unit 42 outputs the identified sound source position information to the part identification unit 44. The process proceeds to step S105, where the sound characteristic acquisition unit 43 analyzes the sound data acquired from the sound data acquisition unit 41 to acquire characteristics of the sound data. The sound characteristic acquisition unit 43 outputs the characteristics of the sound data to the part identification unit 44. The process proceeds to step S107, where the part identification unit 44 identifies the part that emitted the sound based on the sound source position information acquired from the sound source position identification unit 42, the characteristics of the sound data acquired from the sound characteristic acquisition unit 43, and the data stored in the storage device 50. The part identification unit 44 outputs information (name, position) relating to the identified part to the abnormality determination unit 45 .

The process proceeds to step S109, where the abnormality determination unit 45 determines whether the sound data of the part identified by the part identification unit 44 is abnormal. If the sound data of the part is abnormal (YES in step S109), the process proceeds to step S111. On the other hand, if the part is normal (NO in step S109), the process ends. In step S111, the image generation unit 46 generates an image 80 (see FIG. 4 ) indicating the position of the part determined to be abnormal. The image generation unit 46 outputs the generated image 80 to the image output unit 47. The process proceeds to step S113, where the image output unit 47 outputs the image 80 generated by the image generation unit 46 to the display 60.

(Action and Effect)
As described above, the abnormal part detection device according to the first embodiment has the following advantages.

The abnormal part detection device includes a storage device 50 that stores the installation positions of parts mounted on the vehicle 10 in association with the sound characteristics of the parts, and a controller 40. The controller 40 acquires sound data emitted by the vehicle 10. The controller 40 identifies the sound source position from which the sound related to the acquired sound data is generated. The controller 40 identifies the part generating the sound based on the identified sound source position and the installation position stored in the storage device 50. The controller 40 determines whether the sound data is abnormal or not based on the sound characteristics corresponding to the identified part. When the controller 40 determines that the sound data is abnormal, it outputs information indicating that an abnormality has occurred. The information indicating that an abnormality has occurred is information indicating the part in which an abnormality has occurred (the name of the part and an image of the part) and the position of the part, as described in the image 80. Although the display 60 has been taken as an output destination of the image 80, this is not limited thereto. The information indicating that an abnormality has occurred may be output via a speaker or may be output by lighting a lamp. According to the abnormal part detection device, even if a large number of parts are densely mounted on the vehicle 10, it is possible to accurately identify the position of the part generating the abnormal sound.

The sound characteristics of the parts stored in the storage device 50 are characteristics when the parts are normal. The controller 40 compares the sound characteristics of the parts stored in the storage device 50 with the sound characteristics of the acquired sound data, and judges whether the sound data is abnormal or not according to the degree of agreement based on the comparison result. This makes it possible to accurately identify the position of the part emitting the abnormal sound.

The sound characteristics of the parts stored in the storage device 50 include a fundamental frequency based on the shape, size, and material of the part, and harmonic characteristics of the fundamental frequency. This makes it possible to accurately identify the location of the part that is emitting the abnormal sound.

The abnormal part detection device further includes a display device (display 60). When the controller 40 determines that the sound data is abnormal, it outputs to the display device the part that emitted the sound related to the sound data (the name of the part and an image of the part) and information indicating the installation position of the part stored in the storage device 50. This allows the user to grasp at a glance the part in which the abnormality has occurred and the position of that part (see FIG. 4).

Second Embodiment
Next, an abnormal part detection system according to a second embodiment will be described with reference to Figures 6 to 8. The second embodiment differs from the first embodiment in that the computer 30 includes a probability calculation unit 48, as shown in Figure 6. Configurations that overlap with those in the first embodiment will be referred to by reference numerals and their description will be omitted. The following description will focus on the differences.

In the first embodiment, the part with the higher degree of match is identified as the part that emitted the sound. However, there may be cases where there is no difference in the degree of match. For example, in the first embodiment, there may be cases where the part that emitted the sound cannot be identified because there is no difference in the degree of match between "part A1" and "part A2."

Therefore, the probability calculation unit 48 calculates the probability that the part that emitted the sound is "part A1" or "part A2". An example of a probability calculation method will be described. The probability calculation unit 48 calculates the probability based on past identification results. "Past identification results" are results of identifying parts that have emitted sounds up to now. "Past identification results" are stored in the storage device 50. The probability calculation unit 48 creates a learning model by linking the name of the part that emitted the sound at that time to past position coordinates, frequency, sound pressure, and each piece of information related to the part (shape, size, material), and generates a machine learning model using the created learning model. As the machine learning algorithm, for example, Naive Bayes is used. However, the present invention is not limited to this, and any algorithm that allows supervised learning may be used.

The probability calculation unit 48 inputs information such as position coordinates, sound pressure, and frequency when multiple parts with similar degrees of match are extracted, as well as information such as shape, dimensions, and material, into the generated machine learning model, and calculates a probability. The probability calculation unit 48 outputs the calculated probability to the part identification unit 44. The part identification unit 44 identifies the part with the highest calculated probability as the part that emitted the sound.

As shown in Fig. 7, the image generating unit 46 may add the probability calculated by the probability calculating unit 48 to an image 80. The calculated probability (95%) is indicated by reference numeral 73. Furthermore, as indicated by reference numeral 74 in Fig. 7, the image generating unit 46 may add graphs of fundamental frequency and harmonic frequency to the image 80. Note that instead of a frequency graph, a graph that visualizes sound information such as a spectrogram may be added.

Next, an example of the operation of the abnormal part detection device (computer 30) according to the second embodiment will be described with reference to the flowchart of Fig. 8. However, since the processes of steps S201 to S205 and 211 to 215 are similar to the processes of steps S101 to S105 and 109 to 113 shown in Fig. 5, their description will be omitted.

In step S207, the probability calculation unit 48 calculates the probability that the part that emitted the sound is "part A1" or "part A2". The probability calculation unit 48 outputs the calculated probability to the part identification unit 44. The process proceeds to step S209, where the part identification unit 44 identifies the part with the highest calculated probability as the part that emitted the sound.

(Action and Effect)
In this way, the abnormal part detection device according to the second embodiment calculates the probability based on the machine learning model, which makes it possible to accurately identify the position of the part that is emitting the abnormal sound even when no difference in the degree of match is observed.

Each of the functions described in the above embodiments may be implemented by one or more processing circuits. A processing circuit includes a programmed processing device, such as a processor including electrical circuitry. A processing circuit also includes devices such as application specific integrated circuits (ASICs) or circuit components arranged to perform the described functions.

As described above, the embodiment of the present invention has been described, but the description and drawings forming a part of this disclosure should not be understood as limiting this invention. Various alternative embodiments, examples and operating techniques will become apparent to those skilled in the art from this disclosure.

In the above embodiment, the computer 30 has been described as being different from the vehicle 10, but this is not limiting. For example, the functions of the computer 30 (for example, the functions indicated by the reference numerals 41 to 47) may be mounted on the vehicle 10. The microphone 20 may or may not be mounted on the vehicle 10. If the vehicle 10 is not mounted with the microphone 20, it is sufficient for the vehicle 10 to have a receiver that receives sound data transmitted from the microphone 20. In other words, the vehicle 10 can also function as an abnormal part detection device. When the vehicle 10 functions as an abnormal part detection device, the display 60 is a display of a navigation device, a meter display, a HUD (Head-Up Display), or the like.

In the above embodiment, it has been described that the normal sound of the part stored in the storage device 50 is compared with the acquired sound data to determine whether the acquired sound data is abnormal or not. However, the method of determining abnormality is not limited to this. The characteristics of the sound data of the part stored in the storage device 50 may be the characteristics of sound data when the part is abnormal. In other words, abnormal sounds may be compared with each other. As a result of the comparison, if the degree of match is greater than a threshold, the sound data is determined to be abnormal, and if the degree of match is equal to or less than the threshold, the sound data is determined to be normal.

In addition, the abnormal part detection system may include a vehicle 10, a sound collector (microphone 20), and an abnormal part detection device (computer 30) equipped with a storage device 50 that stores the installation positions of parts mounted on the vehicle 10 in association with the sound characteristics of the parts, and a controller 40.

REFERENCE SIGNS LIST 10 Vehicle 20 Microphone 30 Computer 40 Controller 41 Sound data acquisition unit 42 Sound source position identification unit 43 Sound characteristic acquisition unit 44 Part identification unit 45 Abnormality determination unit 46 Image generation unit 47 Image output unit 48 Probability calculation unit 50 Storage device 60 Display

Claims (8)

a storage device that stores, for each part mounted on a vehicle, an installation position of the part and a sound characteristic of the part in association with each other;
A controller.
The controller:
Acquire sound characteristics of sound data emitted by the vehicle;
Detecting a plurality of components having sound characteristics similar to the sound characteristics of the acquired sound data, and comparing installation positions of the detected plurality of components to identify the component that generates the sound;
determining whether the sound data is abnormal based on sound characteristics corresponding to the identified part;
If the sound data is determined to be abnormal, outputting information indicating that an abnormality has occurred;
an abnormal part detection device, characterized in that the information indicating the occurrence of the abnormality is an image simulating an arrangement structure of a plurality of parts, including information indicating the position of the part for which the sound data is determined to be abnormal. the sound characteristics of the component stored in the storage device are characteristics of the component when it is normal,
2. The abnormal part detection device according to claim 1, wherein the controller compares the sound characteristics of the part stored in the storage device with the sound characteristics of the acquired sound data, and determines whether the sound data is abnormal or not depending on a degree of agreement based on a comparison result. 3. The abnormal part detection device according to claim 1, wherein the sound characteristics of the parts stored in the storage device include a fundamental frequency based on the shape, dimensions, and material of the parts, and harmonic characteristics of the fundamental frequency. Further comprising a display device,
The abnormal part detection device according to any one of claims 1 to 3, characterized in that, when the sound data is determined to be abnormal, the controller outputs to the display device information indicating the part that emitted the sound related to the sound data and the installation position of the part stored in the storage device. the sound characteristics of the part stored in the storage device are characteristics when the part is abnormal,
2. The abnormal part detection device according to claim 1, wherein the controller compares the sound characteristics of the part stored in the storage device with the sound characteristics of the acquired sound data, and determines whether the sound data is abnormal or not depending on a degree of agreement based on a comparison result. An abnormal part detection method for an abnormal part detection device including a storage device that stores, for each part mounted on a vehicle, an installation position of the part in association with a sound characteristic of the part, and a controller, comprising:
The controller:
Acquire sound characteristics of sound data emitted by the vehicle;
Detecting a plurality of components having sound characteristics similar to the sound characteristics of the acquired sound data, and comparing installation positions of the detected plurality of components to identify the component that generates the sound;
determining whether the sound data is abnormal based on sound characteristics corresponding to the identified part;
If the sound data is determined to be abnormal, outputting information indicating that an abnormality has occurred;
an abnormal part detection method, characterized in that the information indicating the occurrence of the abnormality is an image simulating an arrangement structure of a plurality of parts, including information indicating a position of the part for which the sound data is determined to be abnormal. An abnormal part detection system including a vehicle, a sound collector, and an abnormal part detection device including a storage device and a controller that stores, for each part mounted on the vehicle, an installation position of the part in association with a sound characteristic of the part,
The controller:
Acquiring sound characteristics of sound data emitted by the vehicle from the sound collector;
Detecting a plurality of components having sound characteristics similar to the sound characteristics of the acquired sound data, and comparing installation positions of the detected plurality of components to identify the component that generates the sound;
determining whether the sound data is abnormal based on sound characteristics corresponding to the identified part;
If the sound data is determined to be abnormal, outputting information indicating that an abnormality has occurred;
an abnormal part detection system, characterized in that the information indicating the occurrence of the abnormality is an image simulating an arrangement structure of a plurality of parts, including information indicating the position of the part for which the sound data is determined to be abnormal. The abnormal part detection device according to any one of claims 1 to 5, characterized in that information indicating the position of a part for which the sound data has been determined to be abnormal is extracted from the layout structure on the image and enlarged, and output together with the image and name of the part for which the sound data has been determined to be abnormal.

Images