Vehicle diagnosis system

Abstract

To provide a vehicle diagnosis system capable of easily performing diagnosis including determination of a type of anomaly using diagnostic sound data.SOLUTION: A vehicle diagnosis system includes a processing circuit 110 and a storage device 120. The storage device 120 stores: data on a learned model; and cluster data generated by inputting, to the learned model, a plurality of pieces of sound data recorded using a plurality of vehicles in which types of anomalies are different from each other and identified, outputting the plurality of pieces of generated data, and clustering, for each of the type of anomalies, a loss variable indicating a magnitude of an error in each of the variables in each piece of the generated data. The processing circuit 110 executes: loss calculation processing that inputs diagnostic sound data of a target vehicle 10 to the learned model and calculates the loss variable indicating the magnitude of the error in each of the variables in the generated data; and diagnostic processing that checks data on the loss variable calculated through the loss calculation processing against the cluster data and outputs a diagnosis result of the target vehicle 10.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] This invention relates to a vehicle diagnosis system.

Background Art

[0002] Patent Document 1 discloses a factor identification system that performs frequency analysis on recorded sound data to identify the factors causing abnormal sounds.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] The factor identification system of Patent Document 1 analyzes the data obtained by Fourier-transforming the sound data to identify the factors causing abnormal sounds. Specifically, the factor identification system divides the Fourier-transformed data set. The factor identification system performs statistical analysis processing on each divided data set to calculate the degree of abnormality. Then, the factor identification system extracts the divided data with a high degree of abnormality as high-abnormality divided data and creates integrated data by integrating the high-abnormality divided data. The factor identification system collates the clusters clustered for each cause of occurrence with the integrated data to identify the cause of the abnormality.

[0005] The factor identification system calculates the probability density for each of the divided data sets. Then, the factor identification system calculates the degree of abnormality using the unconstrained least-squares density ratio estimation method from the calculated probability density. Therefore, the factor identification system has to perform statistical analysis processing repeatedly many times.

[0006] There is a demand for a diagnosis system that can more easily perform a diagnosis including discrimination of the type of abnormality.

Means for Solving the Problem

[0007] Means for solving the above problems and their effects will be described below. The vehicle diagnostic system for solving the above problems diagnoses the target vehicle using sound data obtained by recording sounds emitted from the target vehicle, which is the vehicle to be diagnosed. This vehicle diagnostic system includes a processing circuit and a storage device. The storage device stores training data including training sound data recorded while operating a reference vehicle without any abnormalities, and operation data consisting of a plurality of variables indicating the operating status of the reference vehicle collected simultaneously with the training sound data. The data of a learned model trained by supervised learning to generate the operation data from the training sound data using the training data is stored. A plurality of sound data recorded using a plurality of vehicles with different types of abnormalities and each type of abnormality being identifiable are input to the learned model, and a plurality of generated data, which is the operation data generated using the learned model, are output. Cluster data created by clustering loss variables indicating the magnitude of the error from each variable in the plurality of operation data collected simultaneously with the plurality of sound data for each variable in each generated data for each type of abnormality is stored. The processing circuit inputs diagnostic sound data recorded while operating the target vehicle to the learned model to output the generated data, and executes a loss calculation process for calculating the loss variable in the generated data. The processing circuit collates the data of the loss variable calculated through the loss calculation process with the cluster data, determines to which cluster the loss variable calculated through the loss calculation process belongs, and executes a diagnostic process for outputting a diagnostic result indicating that an abnormality of the type corresponding to the determined cluster has occurred in the target vehicle.

Advantages of the Invention

[0008] According to the above vehicle diagnostic system, by using a learned model, it is possible to easily perform a diagnosis including discrimination of the type of abnormality using diagnostic sound data.

Brief Description of the Drawings

[0009]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

DETAILED DESCRIPTION OF THE INVENTION

[0010] Hereinafter, an embodiment of the vehicle diagnostic system will be described with reference to FIGS. 1 to 6. <Configuration of Vehicle Diagnostic System> As shown in FIG. 1, this vehicle diagnostic system includes a data center 100 and a data acquisition device 300. The data center 100 is communicably connected to the data acquisition device 300 via a communication network 200. As shown in FIG. 1, the data center 100 includes a storage device 120 in which a program is stored, and a processing circuit 110. The processing circuit 110 executes various processes by executing the program stored in the storage device 120. The data center 100 includes a communication device 130.

[0011] The data acquisition device 300 is, for example, a personal computer. The data acquisition device 300 includes a storage device 320 in which a program is stored, and a processing circuit 310. The processing circuit 310 executes various processes by executing the program stored in the storage device 320. The data acquisition device 300 includes a communication device 330. In this embodiment, the data acquisition device 300 is connected to the data center 100 by wireless communication via the communication network 200. The data acquisition device 300 includes a display device 340 for displaying information. The data acquisition device 300 includes a microphone 350.

[0012] When diagnosing the target vehicle 10, which is the vehicle to be diagnosed, using this vehicle diagnostic system, the microphone 350 is installed at a predetermined position with respect to the target vehicle 10. The data acquisition device 300 is connected to the vehicle control unit 20 of the target vehicle 10. Then, an operator operates the target vehicle 10 to operate the target vehicle 10 in a measurement driving pattern. The measurement driving pattern is a driving pattern suitable for acquiring data for diagnosis and is a predetermined driving pattern over a predetermined time. When the target vehicle 10 is operating in the measurement driving pattern in this way, the data acquisition device 300 records sound with the microphone 350. The data acquisition device 300 acquires operation data indicating the operation status of the target vehicle 10 simultaneously with the recording of the sound data.

[0013] The vehicle control unit 20 controls each part of the target vehicle 10. Various sensors for detecting the state of the target vehicle 10 are connected to the vehicle control unit 20. When the data acquisition device 300 is connected to the vehicle control unit 20, the data acquisition device 300 can acquire information on the target vehicle 10 through the vehicle control unit 20.

[0014] <Overview of Diagnosis by Vehicle Diagnostic System> As described above, in this vehicle diagnostic system, when diagnosing the target vehicle 10, the data acquisition device 300 is connected to the vehicle control unit 20 of the target vehicle 10. Then, while the target vehicle 10 is in operation, the data acquisition device 300 records sounds with the microphone 350. The data acquisition device 300 transmits data including the recorded sound data to the data center 100. Then, the data center 100 diagnoses the target vehicle 10 using the received data.

[0015] The data acquisition device 300 records, as diagnostic sound data, the sound data recorded by the microphone 350 while operating the target vehicle 10 in a measurement driving pattern over a preset time period in the storage device 320. The data acquisition device 300 stores the operation data collected simultaneously with the sound data in the storage device 320. For example, when diagnosing the transmission of the target vehicle 10, the operation data is the engine rotational speed NE, the input rotational speed Nin, the output rotational speed Nout, and the gear ratio. Then, the data acquisition device 300 stores the operation data including the thus collected diagnostic sound data in the storage device 320 as a data set for a preset time period.

[0016] The data acquisition device 300 extracts a data set for a preset time stored in the storage device 320 for each piece of data within the range of a window Tw with a time width shorter than the preset time while changing the extraction start time, and formats it into diagnostic data. That is, the data acquisition device 300 extracts data for a period shorter than the preset time and formats it into diagnostic data. In the data formatting process for formatting the diagnostic data, the data acquisition device 300 converts the diagnostic audio data into a mel spectrogram and treats it as image data. The vertical axis of the mel spectrogram is the frequency indicated by the mel scale. And the horizontal axis is the time axis. In the mel spectrogram, the intensity is represented by color. The lower the intensity, the darker the blue color, and the higher the intensity, the brighter the red color. The audio data included in one data set for the preset time becomes one mel spectrogram for the preset time. The data acquisition device 300 transmits the formatted diagnostic data to the data center 100. When receiving the diagnostic data, the data center 100 stores the diagnostic data in the storage device 120. And the data center 100 executes the routine shown in FIG. 2 to diagnose the target vehicle 10. This routine is executed by the processing circuit 110 of the data center 100. In the following description, the step numbers of each process are indicated by attaching "S" before the numbers.

[0017] As shown in FIG. 2, the data center 100 executes a loss calculation process (S100) and a diagnosis process (S200) to diagnose the target vehicle 10. The loss calculation process (S100) is a process of inputting the diagnostic audio data included in the diagnostic data into a learned model to calculate a loss variable LOS. The loss variable LOS is a value indicating the magnitude of the error of the generated data generated by the learned model.

[0018] The learned model is a model trained by supervised learning so that it can generate operation data from training sound data using a large amount of training data including the training sound data and the operation data collected simultaneously with the training sound data. For example, in the data center 100, diagnostic sound data is treated as image data. Therefore, the data center 100 uses a model that partially uses ResNet-18, which is an image classification model. ResNet-18 is a pre-trained image classification model trained on the ImageNet dataset. ResNet-18 has been trained with over 1 million image data and can classify the input images into 1000 categories. The learned model stored in the storage device 120 of the data center 100 is a model obtained by transfer learning of the pre-trained ResNet-18. This learned model is a model in which the output layer for classification of ResNet-18 is replaced with a neural network MLP, and this neural network MLP is trained by supervised learning. Training means adjusting the weights and biases of each layer of this neural network MLP. The training sound data is sound data recorded while operating a reference vehicle in a state serving as a diagnostic standard according to a measurement driving pattern. In this example, a vehicle that has completed a certain running-in operation after manufacture, undergone sufficient maintenance inspections, and been confirmed to have no abnormalities is used as the reference vehicle.

[0019] The learned model is optimized to generate operation data from the sounds emitted by the reference vehicle. Therefore, when diagnostic sound data emitted from the target vehicle 10 in a state different from that of the reference vehicle is input, the operation status data cannot be correctly generated. That is, if the state of the target vehicle 10 deviates from the state of the reference vehicle, a deviation occurs between the operation data stored in the data set as correct data and the generated data. A large deviation indicates that the state of the target vehicle 10 deviates from the state of the reference vehicle by that much. As described above, the reference vehicle is a vehicle in a normal state. Therefore, in this vehicle diagnosis system, the abnormality of the target vehicle 10 is diagnosed based on this deviation, that is, the error of the generated data. The way the error appears varies depending on the type of abnormality.

[0020] FIG. 3 is a graph showing the operation data and the generated data generated using the learned model. In the graph of FIG. 3, only one of the plurality of variables included in the operation data is shown. That is, only one of the target variables that are the outputs of the learned model is shown. In FIG. 3, the operation data which is the correct data is shown by a solid line. In FIG. 3, the generated data is shown by a square symbol. The time ts in FIG. 3 is the start time of the measurement operation pattern, and the time tf is the end time of the measurement operation pattern. That is, FIG. 3 shows the operation data and the generated data in one data set. The learned model outputs the value of the generated data in the window Tw for each variable one by one based on the data in the range of the window Tw shaped into a list. The error ERR indicates the deviation of the generated data from the operation data which is the correct data, that is, the error. The greater the deviation of the state of the target vehicle 10 from the state of the reference vehicle, the greater the error ERR. Therefore, this vehicle diagnosis system calculates a loss variable LOS which is the sum of the errors ERR for each variable included in the operation data in one data set through a loss calculation process. The loss variable LOS is a set of the sums of the errors ERR in one data set for each of the variables included in the operation data. Therefore, the loss variable LOS is a vector having the same number of dimensions as the number of variables included in the operation data. For example, when the variables included in the operation data are two, the first variable and the second variable, the loss variable LOS is a two-dimensional vector consisting of the first loss variable LOS1 and the second loss variable LOS2. The loss variable LOS only needs to indicate the magnitude of the error. Therefore, the loss variable LOS does not have to be a simple sum of the errors ERR. For example, as a preprocess for calculating the loss variable LOS, a process of excluding outliers from the plurality of errors ERR may be performed, and the sum after excluding the outliers may be used as the loss variable LOS. Alternatively, the median of the errors ERR may be used as the loss variable LOS.

[0021] The storage device 120 of the data center 100 stores cluster data created using a plurality of sound data recorded using vehicles for which the type of abnormality can be specified. As described above, the way the error appears differs depending on the type of abnormality.

[0022] Figure 4 shows the distribution of the loss variable LOS in the cluster data. The vertical axis in Figure 4 indicates the magnitude of the first loss variable LOS1. The first loss variable LOS1 is the sum of the errors ERR for the first variable of the operation data. And the horizontal axis in Figure 4 indicates the magnitude of the second loss variable LOS2. The second loss variable LOS2 is the sum of the errors ERR for the second variable of the operation data.

[0023] Figure 4 shows the cluster data created using three test vehicles for which the types of anomalies are each specified. The anomalies of the three test vehicles are different from each other. This cluster data is created using a plurality of sound data recorded by repeatedly performing test runs in a measurement driving pattern a plurality of times for each of the three test vehicles. The cluster data is data created by clustering a plurality of loss variables LOS calculated by executing the same processing as the loss calculation processing using the plurality of sound data thus recorded, for each type of anomaly.

[0024] In Figure 4, the coordinates of the loss variable LOS classified into the first cluster, which is the cluster corresponding to the first test vehicle in which the first anomaly has occurred, are shown by white triangles. In Figure 4, the coordinates of the loss variable LOS classified into the second cluster, which is the cluster corresponding to the second test vehicle in which the second anomaly has occurred, are shown by white circles. In Figure 4, the coordinates of the loss variable LOS classified into the third cluster, which is the cluster corresponding to the third test vehicle in which the third anomaly has occurred, are shown by white squares. And in Figure 4, the center of gravity of each cluster is shown by a cross mark. The center of gravity C1 is the center of gravity of the first cluster. The center of gravity C2 is the center of gravity of the second cluster. The center of gravity C3 is the center of gravity of the third cluster.

[0025] In the storage device 120 of the data center 100, data of the center of gravity of each cluster clustered by type of abnormality is stored as cluster data in this way. FIG. 4 shows two examples of the variables included in the loss variable LOS. The number of variables in the loss variable LOS is equal to the number of variables included in the operation data. That is, FIG. 4 is an example of a space in which coordinates are defined by the magnitude of the error of each variable included in the operation data. The number of clusters to be classified, that is, the number of types of abnormalities to be diagnosed, is not limited to three. It may be two or four or more. Prepare a test vehicle according to the type of abnormality to be diagnosed, calculate the loss variable LOS, and create cluster data.

[0026] This vehicle diagnosis system performs a diagnosis process using the cluster data stored in the storage device 120, that is, the data of the center of gravity of each cluster. Hereinafter, the content of the loss calculation process and the diagnosis process will be described with reference to a flowchart.

[0027] <Regarding the loss calculation process> As described above, when receiving the diagnostic data, the data center 100 stores the diagnostic data in the storage device 120. Then, the processing circuit 110 of the data center 100 executes the routine shown in FIG. 2.

[0028] As shown in FIG. 2, when starting this routine, the processing circuit 110 first executes a loss calculation process (S100). FIG. 5 is a flowchart showing the flow of processing in the loss calculation process. When this routine is started, the processing circuit 110 reads diagnostic data (S110). Then, the processing circuit 110 sequentially inputs a plurality of lists included in the read diagnostic data into the learned model to calculate data on the operating status respectively (S120). That is, the processing circuit 110 generates a plurality of generated data. Next, the processing circuit 110 calculates the magnitude of the error ERR in the generated data for each variable in the operating data collected simultaneously with the diagnostic sound data for all the generated data (S130). Then, the processing circuit 110 stores and records the calculated error ERR in the storage device 120 (S130).

[0029] Next, the processing circuit 110 calculates a loss variable LOS (S140). Then, the processing circuit 110 stores and records the calculated loss variable LOS in the storage device 120 (S140). When the loss variable LOS is recorded, the processing circuit 110 ends this routine.

[0030] As shown in FIG. 2, when the loss calculation process (S100) ends, the processing circuit 110 executes a diagnosis process (S200). <Regarding the diagnosis process> FIG. 6 is a flowchart showing the flow of processing in the diagnosis routine. When this routine is started, the processing circuit 110 reads the data of the loss variable LOS recorded in the storage device 120 (S210).

[0031] Next, the processing circuit 110 collates the cluster data recorded in the storage device 120 with the loss variable LOS to determine which cluster the loss variable LOS belongs to (S220). For example, the processing circuit 110 calculates the distance between the loss variable LOS and the centroid of each cluster. Then, the processing circuit 110 determines that the loss variable LOS belongs to the cluster having the centroid with the closest distance. The distance may be calculated using any distance calculation method such as Euclidean distance, Mahalanobis distance, Manhattan distance. A calculation method suitable for diagnosis may be adopted.

[0032] In FIG. 4, the coordinates of the loss variable LOS calculated through the loss calculation process are indicated by the symbols of black-filled squares. In the case of the example shown in FIG. 4, the centroid closest to the coordinates of the loss variable LOS is the centroid C3. Therefore, in the case of this example, the processing circuit 110 determines that the loss variable LOS belongs to the third cluster.

[0033] When the distance from the origin where the error of each variable is 0 to the coordinates of the loss variable LOS is within a predetermined distance, the processing circuit 110 determines that no abnormality has occurred in the target vehicle 10. Based on the fact that the distance is within the predetermined distance, the magnitude of the predetermined distance is set to such a size that it can be determined that the coordinates of the loss variable LOS are sufficiently close to the origin, so that the state of the target vehicle 10 is close to the state of the reference vehicle and no abnormality has occurred.

[0034] Next, the processing circuit 110 outputs a diagnostic result corresponding to the determination result (S230). Specifically, the processing circuit 110 outputs a diagnostic result indicating that an abnormality of the type corresponding to the cluster to which the loss variable LOS belongs has occurred in the target vehicle 10. For example, in the case of the example shown in FIG. 4, the processing circuit 110 outputs a diagnostic result indicating that the third abnormality corresponding to the third cluster has occurred. When it is determined that no abnormality has occurred in the target vehicle 10, the processing circuit 110 outputs a diagnostic result indicating that the target vehicle 10 is normal.

[0035] After outputting the diagnostic result in this way, the processing circuit 110 ends this routine. That is, the processing circuit 110 ends the diagnostic process and ends the series of routines shown in FIG. 2.

[0036] The data acquisition device 300 that has received the diagnostic result displays the received diagnostic result on the display device 340 as the diagnostic result of the target vehicle 10. <Actions of this Embodiment> The data center 100 generates generated data by restoring operation data from diagnostic data using a learned model. Then, the data center 100 executes a loss calculation process (S100) for calculating a loss variable LOS in the generated data.

[0037] In the diagnosis process (S200), the data center 100 collates the coordinates of the loss variable LOS with the cluster data. Then, it diagnoses that an abnormality of the type corresponding to the cluster to which the loss variable LOS belongs has occurred.

[0038] The vehicle diagnosis system uses the generated data calculated using a learned model and the loss variable LOS calculated by comparing with the operation data. The difference in the states of the target vehicle 10 and the reference vehicle appears in the loss variable LOS. The characteristics corresponding to the type of abnormality occurring in the target vehicle 10 also appear in the loss variable LOS. Therefore, by collating the loss variable LOS of the target vehicle 10 with the data of a plurality of clusters created using the loss variable LOS of a plurality of vehicles for which the type of abnormality has been specified, the type of abnormality can be specified.

[0039] <Advantages of this Embodiment> (1) According to the vehicle diagnosis system, by using a learned model, it is possible to easily perform a diagnosis including discrimination of the type of abnormality using diagnostic sound data.

[0040] (2) When the distance from the origin to the coordinates of the loss variable LOS is within a predetermined distance, in the diagnosis process, the processing circuit 110 outputs a diagnosis result indicating that no abnormality has occurred in the target vehicle 10. The closer the state of the target vehicle 10 is to the reference vehicle, the smaller the error between the generated data and the operation data. That is, when no abnormality has occurred in the target vehicle 10, the coordinates of the loss variable LOS are close to the origin. According to the above vehicle diagnosis system, based on the fact that the distance from the origin to the coordinates of the loss variable LOS is within a predetermined distance, it is possible to diagnose that no abnormality has occurred in the target vehicle 10.

[0041] <Modification Example> This embodiment can be implemented with the following modifications. This embodiment and the following modification examples can be implemented in combination with each other within a technically consistent range.

[0042] · The data used as diagnostic data does not have to be a Mel spectrogram. For example, a spectrogram obtained by wavelet-transforming audio data may be used. A spectrogram obtained by performing a short-time Fourier transform on audio data may also be used. It is not essential to convert audio data into image data. For example, feature amounts may be extracted from the audio data itself and used as diagnostic data. In that case, it is not necessary to use ResNet-18, which handles image data, as the learned model. Although a model obtained by transfer learning of ResNet-18 was exemplified, the structure of the model is not limited to such a structure. The learned model only needs to be able to output generated data based on the diagnostic data.

[0043] · In the above embodiment, the data of the centroid of each cluster was stored in the storage device 120 as the cluster data. The cluster data is not limited to the data of the centroid. The cluster data may be any data that can be collated with the data of the loss variable LOS calculated through the loss calculation process to determine which cluster the loss variable LOS belongs to. For example, it may be data storing the values of all the loss variables constituting each cluster.

[0044] · The vehicle diagnostic system may be composed only of the data acquisition device 300. In this case, the data of the learned model and the cluster data are stored in the storage device 320. In this case, the processing circuit 310 of the data acquisition device 300 executes the loss calculation process and the diagnostic process.

[0045] · The data acquisition device 300 itself does not have to be equipped with the microphone 350. It is also possible to acquire audio data from an external device and perform the loss calculation process. The loss calculation process may be performed using a plurality of audio data recorded using a plurality of microphones 350.

[0046] ·In the diagnosis process, when the processing circuit 110 determines that it does not belong to any cluster, the target vehicle 10 may be output with a diagnosis result indicating that an abnormality other than the abnormalities corresponding to each cluster has occurred.

[0047] In this case, when any of the distances between the coordinates of the loss variable LOS and the centroids of the respective clusters exceeds the distance set as the upper limit, the processing circuit 110 outputs a diagnosis result to the target vehicle 10 indicating that an abnormality other than the abnormalities corresponding to each cluster has occurred.

[0048] Based on the fact that the distance between the centroid of each cluster and the loss variable LOS calculated through the error output process exceeds the upper limit threshold, it is determined that it does not belong to any cluster. Thereby, it can be determined that an abnormality other than the abnormality corresponding to each cluster has occurred. The upper limit distance may be set as the upper limit distance of the distance at which the loss variable LOS is determined to belong to the cluster.

[0049] ·The operation for acquiring diagnostic data may be performed a plurality of times, and a plurality of loss variables LOS may be calculated for diagnosis. For example, the average value of the plurality of loss variables LOS may be calculated and the diagnosis process may be performed using the average value.

[0050] ·In the above embodiment, as an example of diagnosing the transmission of the target vehicle 10, an example is shown in which the operation data includes data on the rotational speed of the rotating shaft in the power train. The vehicle diagnostic system may perform the diagnosis of the target vehicle 10 by diagnosing other units in the target vehicle 10. For example, in order to diagnose the engine, it is conceivable to use data on the engine rotational speed NE, ignition timing, and engine load ratio as the operation data. For example, in order to diagnose the drive motor, it is conceivable to use data on the rotational speed of the output shaft of the motor as the operation data. In order to diagnose a four-wheel drive vehicle, it is conceivable to use data on the rotational speed of each drive wheel as the operation data.

[0051] · In the above embodiment, the vehicle diagnostic system includes a processing circuit 110 and a storage device 120, and executes software processing. However, this is merely an example. For example, the vehicle diagnostic system may include a dedicated hardware circuit (such as an ASIC, etc.) that processes at least a part of the software processing executed in the above embodiment. That is, the vehicle diagnostic system may have any of the following configurations (A) to (C). (A) The vehicle diagnostic system includes an execution device that executes all processing according to a program, and a storage device that stores the program. That is, the vehicle diagnostic system includes a software execution device. (B) The vehicle diagnostic system includes an execution device that executes a part of the processing according to a program, and a storage device. Further, the vehicle diagnostic system includes a dedicated hardware circuit that executes the remaining processing. (C) The vehicle diagnostic system includes a dedicated hardware circuit that executes all processing. There may be a plurality of software execution devices and / or dedicated hardware circuits. That is, the above processing may be executed by a processing circuitry including at least one of one or more software execution devices and one or more dedicated hardware circuits. The storage device, that is, the computer-readable medium, storing the program includes any available medium accessible by a general-purpose or dedicated computer.

Explanation of Reference Numerals

[0052] 10… Target vehicle, 20… Vehicle control unit, 100… Data center, 110… Processing circuit, 120… Storage device, 130… Communication device, 200… Communication network, 300… Data acquisition device, 310… Processing circuit, 320… Storage device, 330… Communication device, 340… Display device, 350… Microphone

Claims

1. A vehicle diagnosis system for diagnosing a target vehicle by using sound data obtained by recording sounds emitted from the target vehicle, which is the vehicle to be diagnosed, comprising a processing circuit and a storage device, data of a trained model trained by supervised learning to generate the operation data from the training sound data by using training data including the training sound data recorded while operating a reference vehicle without any abnormality and operation data consisting of a plurality of variables indicating the operation status of the reference vehicle collected simultaneously with the training sound data, a plurality of generated data which are operation data generated by using the learned model by inputting a plurality of sound data recorded by using a plurality of vehicles having different types of abnormalities and each having a specific type of abnormality into the learned model, and loss variables indicating the magnitude of errors from the respective variables in the plurality of operation data collected simultaneously with the plurality of sound data for each of the respective variables in each of the generated data, and cluster data created by clustering for each type of abnormality, which are stored in the storage device, a loss calculation process for inputting the diagnostic sound data recorded while operating the target vehicle into the learned model to output the generated data and calculating the loss variable in the generated data, a diagnostic process for collating the data of the loss variable calculated through the loss calculation process with the cluster data to determine to which cluster the loss variable calculated through the loss calculation process belongs, and outputting a diagnostic result indicating that an abnormality of the type corresponding to the determined cluster has occurred in the target vehicle, wherein the processing circuit executes the vehicle diagnosis system.

2. In the diagnostic process, the processing circuit calculates the distance between the coordinates of the loss variable in a space defined by the magnitude of the error for each of the variables included in the operation data and the centroid of each of the clusters in the space, and determines that the loss variable belongs to the cluster having the centroid closest to the coordinates of the loss variable. The vehicle diagnosis system according to Claim 1.

3. In the diagnostic process, when any of the distances between the coordinates of the loss variable and the centroids of the respective clusters exceeds a distance set as an upper limit, the processing circuit outputs a diagnostic result indicating that an abnormality other than the abnormalities corresponding to the respective clusters has occurred in the target vehicle. The vehicle diagnosis system according to claim 2.

4. When the distance from the origin where the errors of the respective variables are all 0 to the coordinates of the loss variable is within a predetermined distance in the space defined by coordinates according to the magnitudes of the errors of the respective variables included in the operation data, in the diagnosis process, the processing circuit outputs a diagnosis result indicating that no abnormality has occurred in the target vehicle. The vehicle diagnosis system according to claim 1.

5. The operation data includes data on the rotational speed of the rotating shaft in the power train. The vehicle diagnosis system according to claim 1.

Images