Vehicle abnormal sound monitoring data processing method, vehicle end, cloud end and vehicle

Abstract

The invention provides a vehicle abnormal sound monitoring data processing method, a vehicle end, a cloud end and a vehicle, and relates to the technical field of vehicle monitoring, and the method comprises the steps: processing the audio monitoring data of the vehicle, and obtaining a sound feature; inputting the sound features into an edge depth self-encoder at the vehicle end, and judging whether the vehicle is abnormal or not; if it is judged that the vehicle is suspected to be abnormal, abnormal monitoring data are determined according to the time point when the vehicle is abnormal, the preset duration and the operation data of the vehicle; the abnormal monitoring data are uploaded to a cloud, and the cloud verifies whether the vehicle is abnormal or not according to the abnormal monitoring data. The vehicle abnormal sound can be monitored locally at the vehicle end according to the edge depth auto-encoder, the data does not need to be uploaded to the cloud end, the local vehicle end can respond to the abnormal sound in time, and the monitoring stability is ensured; the vehicle end and the cloud end achieve a cloud side synergistic effect, and timeliness, stability and accuracy of monitoring are considered.

Description

Technical Field

[0001] The present invention relates to the field of vehicle monitoring technology, and in particular to a vehicle abnormal noise monitoring data processing method, a vehicle terminal, a cloud and a vehicle. Background Art

[0002] Vehicle electronic and electrical architecture is becoming increasingly complex, and modern vehicles have an urgent need for remote monitoring and management. By real-time monitoring of the vehicle's operating status, users can prevent vehicle failures and reduce maintenance costs. In the existing technology, in the process of monitoring vehicle abnormal noises, the collected data is uploaded to the cloud in real time, the collected vehicle data is processed in the cloud, and the processed results are transmitted to the vehicle side. However, the amount of data collected in real time is large, the data transmission capacity between the vehicle side and the cloud side is limited, and data transmission is congested, resulting in a delay in the vehicle side's response to abnormal noise failures and reduced real-time performance. Summary of the invention

[0003] The problem to be solved by the present invention is that the amount of data transmitted between the vehicle side and the cloud in the prior art is large, which causes the vehicle side to respond to abnormal noise failures lagging and reduces real-time performance.

[0004] In order to solve the above problems, in a first aspect, the present invention provides a vehicle abnormal noise monitoring data processing method, which is applied to a vehicle end and comprises:

[0005] Processing the audio monitoring data of the vehicle to obtain sound characteristics;

[0006] Inputting the sound feature into an edge depth autoencoder to determine whether the vehicle is abnormal, wherein the edge depth autoencoder is arranged at the vehicle end;

[0007] If it is determined that the vehicle is suspected of having an abnormality, determining abnormal monitoring data according to the time point when the vehicle has an abnormality, a preset duration, and the operating data of the vehicle;

[0008] The abnormal monitoring data is uploaded to the cloud, wherein the cloud is used to verify whether the vehicle has an abnormality based on the abnormal monitoring data.

[0009] Optionally, inputting the sound feature into an edge deep autoencoder to determine whether the vehicle is abnormal includes:

[0010] Inputting the sound feature into an edge deep autoencoder to obtain a sound reconstruction feature, wherein the edge deep autoencoder is obtained by the cloud sending a deep autoencoder trained with a training set to the vehicle end, wherein the training set includes normal audio monitoring data;

[0011] Obtaining a loss value according to the sound reconstruction feature and the sound feature;

[0012] Determining whether the loss value is greater than a reconstruction error threshold;

[0013] If the loss value is greater than the reconstruction error threshold, it is determined that the vehicle is suspected of being abnormal.

[0014] Optionally, if it is determined that the vehicle is suspected of having an abnormality, determining the abnormal monitoring data according to the time point when the vehicle has an abnormality, a preset duration, and the operation data of the vehicle includes:

[0015] If it is determined that the vehicle is suspected of having an abnormality, determining an abnormal time period according to the abnormal time point and the preset duration;

[0016] Marking the audio monitoring data and the vehicle operation data corresponding to the abnormal time period as suspected abnormal audio monitoring data and suspected abnormal vehicle operation data, respectively;

[0017] Determining whether the suspected abnormal vehicle operation data meets a preset abnormality rule;

[0018] If the suspected abnormal vehicle operation data satisfies the preset abnormal rule, the abnormal monitoring data is determined, wherein the abnormal monitoring data includes the suspected abnormal audio monitoring data and the suspected abnormal vehicle operation data.

[0019] Optionally, the suspected abnormal vehicle operation data includes fault codes and operation control parameters;

[0020] The determining whether the suspected abnormal vehicle operation data satisfies a preset abnormality rule includes:

[0021] Inputting the operation control parameters corresponding to the various components of the vehicle into the corresponding component diagnosis model to diagnose whether each of the components is operating abnormally;

[0022] Obtaining a total number of abnormalities according to the number of the fault codes in the suspected abnormal vehicle operation data and the number of the components diagnosed with abnormal operation;

[0023] Determine whether the total number of abnormalities is greater than a preset number threshold;

[0024] If the total number of anomalies is greater than the preset number threshold, the suspected abnormal vehicle operation data satisfies the preset abnormality rule.

[0025] Optionally, before inputting the sound feature into the edge deep autoencoder to determine whether the vehicle is abnormal, the vehicle abnormal sound monitoring data processing method further includes:

[0026] According to the deep autoencoder periodically sent down by the cloud, the edge deep autoencoder is updated, wherein the cloud is used to extract the sound features of the normal audio monitoring data in the training set, input the sound features into the deep autoencoder, obtain the reconstructed sound features, and compare the sound features with the reconstructed sound features; if the loss value between the reconstructed sound features and the sound features is greater than or equal to the allowable loss value, adjust the parameters of the deep autoencoder and continue training; if the loss value between the reconstructed sound features and the sound features is less than the allowable loss value, stop training, and obtain the trained deep autoencoder, and the training set includes the normal audio monitoring data obtained from the vehicle end.

[0027] Optionally, before inputting the sound feature into the edge deep autoencoder to determine whether the vehicle is abnormal, the vehicle abnormal sound monitoring data processing method further includes:

[0028] The reconstruction error threshold of the vehicle is updated according to the reconstruction error threshold periodically sent down by the cloud, wherein the reconstruction error threshold periodically sent down by the cloud is greater than a preset proportion of the loss values ​​in a loss value set, and the loss value set is obtained by the loss value of the deep autoencoder corresponding to each normal audio monitoring data in the cloud statistical test set.

[0029] Optionally, before uploading the abnormal monitoring data to the cloud, the vehicle abnormal noise monitoring data processing method further includes:

[0030] Determine whether user authorization information is received;

[0031] When the user authorization information is received and the abnormal monitoring data is desensitized, the abnormal monitoring data is uploaded to the cloud.

[0032] In a second aspect, the present invention further provides a vehicle end, comprising:

[0033] A data processing module, used to process the audio monitoring data of the vehicle to obtain sound characteristics;

[0034] The edge diagnosis module is used to input the sound features into the edge deep autoencoder to determine whether the vehicle has an abnormality, wherein the edge deep autoencoder is arranged on the vehicle side; if it is determined that the vehicle is suspected of having an abnormality, the abnormal monitoring data is determined according to the time point when the vehicle has an abnormality, a preset duration and the operation data of the vehicle.

[0035] The main control module is used to upload the abnormal monitoring data to the cloud, wherein the cloud is used to verify whether the vehicle has an abnormality based on the abnormal monitoring data.

[0036] In a third aspect, a vehicle abnormal noise monitoring data processing method is applied to the cloud, comprising:

[0037] According to the abnormal monitoring data sent by the vehicle side, check whether the vehicle has any abnormality, wherein the abnormal monitoring data is processed by the vehicle side on the audio monitoring data of the vehicle to obtain sound features, and the sound features are input into the edge deep autoencoder to determine whether the vehicle has any abnormality. If it is determined that the vehicle is suspected of having an abnormality, it is determined based on the time point when the vehicle has the abnormality, the preset duration and the operating data of the vehicle.

[0038] In a fourth aspect, the present invention further provides a cloud, comprising:

[0039] The data verification module is used to verify whether the vehicle has an abnormality based on the abnormal monitoring data sent by the vehicle side, wherein the abnormal monitoring data is obtained by processing the vehicle's audio monitoring data by the vehicle side to obtain sound features; the sound features are input into the edge deep autoencoder to determine whether the vehicle has an abnormality; if it is determined that the vehicle is suspected of having an abnormality, it is determined based on the time point of the vehicle abnormality, the preset duration and the vehicle's operating data.

[0040] In a fifth aspect, the present invention further provides a vehicle, comprising a memory and a processor, wherein the memory is used to store a computer program, and the processor is used to implement the vehicle abnormal sound monitoring data processing method described in the first aspect or the vehicle abnormal sound monitoring data processing method described in the third aspect when executing the computer program.

[0041] Compared with the prior art, the present invention has the following beneficial effects:

[0042] The present invention provides a vehicle abnormal noise monitoring data processing method, a vehicle side, a cloud and a vehicle. The vehicle audio monitoring data is processed to obtain sound features, and then the sound features are input into the edge deep autoencoder of the vehicle side to determine whether the vehicle is abnormal. The vehicle abnormal noise can be monitored locally on the vehicle side according to the edge deep autoencoder, and the data does not need to be uploaded to the cloud. The data will not be delayed due to the cloud transmission problem. The local vehicle side can respond to the abnormality in time to ensure the stability of monitoring. If it is determined that the vehicle is suspected of being abnormal, the abnormal monitoring data is determined according to the time point of the abnormality, the preset duration and the operation data of the vehicle, and the abnormal monitoring data is further located. The time point of the abnormality and the monitoring data of the vehicle when the abnormal noise occurs are determined, which is convenient for subsequent troubleshooting and analysis of the abnormal noise fault and improves the rate of fault resolution. The abnormal monitoring data is uploaded to the cloud, wherein the cloud verifies whether the vehicle is abnormal according to the abnormal monitoring data, for example, using the latest deep autoencoder stored in the cloud for verification. After multiple verifications, the accuracy of monitoring is improved, and the vehicle side and the cloud side achieve a cloud-edge synergy effect, taking into account the timeliness, stability and accuracy of monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A schematic flow chart of a vehicle abnormal noise monitoring data processing method provided by an embodiment of the present invention;

[0044] Figure 2 A schematic flow chart of another vehicle abnormal noise monitoring data processing method provided by an embodiment of the present invention;

[0045] Figure 3 A schematic diagram of the process of training a deep autoencoder in the cloud provided by an embodiment of the present invention;

[0046] Figure 4 A schematic diagram of the structure of the vehicle side and the cloud side provided by an embodiment of the present invention;

[0047] Figure 5 A schematic structural diagram of a vehicle provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0048] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0049] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein.

[0050] In the description of this specification, the description with reference to the terms "embodiment", "one embodiment" and "one implementation" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or implementation are included in at least one embodiment or implementation of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or implementation. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or implementations in a suitable manner.

[0051] like Figure 1 As shown, a vehicle abnormal noise monitoring data processing method provided by an embodiment of the present invention is applied to a vehicle side, comprising:

[0052] S1: Process the audio monitoring data of the vehicle to obtain sound features.

[0053] Specifically, a sound collection device, such as a sound collection box or a recording box, can be installed inside the vehicle to collect the sound emitted by the vehicle. In addition, the position and number of the installed sound collection devices can be adjusted as needed. For example, the sound collection devices can be installed at the front, middle and rear positions of the vehicle, or near multiple components of particular concern in the vehicle (such as the engine, drive motor 1, drive motor 2 and transmission, etc.). The sound collection device is connected to the vehicle end through a signal transmission line, so that the vehicle end can obtain the audio monitoring data of the vehicle; the sound collection device can periodically upload the audio monitoring data. Features are extracted from the audio monitoring data through spectrum analysis. The audio feature extraction method includes: the audio monitoring data can be converted into a spectrogram, and then the spectrum centroid, spectrum roll-off, spectrum flatness or sampling frequency, etc. can be extracted from the spectrum. The spectrum can also be mapped to the Mel frequency to obtain the Mel spectrum. The Mel spectrum is processed and feature extracted to obtain the Mel frequency energy coefficient or the Mel frequency cepstrum coefficient. Select one of the above feature extraction methods to extract the sound features of the original audio clip, or obtain a sound feature combination after splicing. The sound feature can refer to the audio of the sound.

[0054] S2: Input the sound feature into an edge depth autoencoder to determine whether the vehicle has an abnormality, wherein the edge depth autoencoder is arranged at the vehicle end.

[0055] Specifically, the edge deep autoencoder can be an encoder pre-set on the vehicle side, or it can be an autoencoder based on LSTM neural network (Long Short-Term Memory), CNN neural network (Convolutional Neural Networks) or other neural networks in the cloud. After training, the deep autoencoder is obtained, and then the cloud sends the deep autoencoder to the vehicle side as the edge deep autoencoder. The edge deep autoencoder is used to reconstruct the sound features, for example, by comparing the output results of the edge deep autoencoder and the input sound features, analyzing the differences between the two, so as to determine whether the vehicle is abnormal; it can also analyze whether the output results are within the normal range, so as to determine whether the vehicle is abnormal, etc.

[0056] S3: If it is determined that the vehicle is suspected of being abnormal, determine abnormal monitoring data according to the time point when the vehicle is abnormal, the preset duration and the operation data of the vehicle. If it is determined that the vehicle is abnormal, continue a new round of diagnostic monitoring.

[0057] Specifically, the vehicle abnormal noise can be monitored locally on the vehicle side based on the edge deep autoencoder. There is no need to upload the data to the cloud. The data will not be delayed due to the cloud, and the local vehicle side can respond to the abnormality in time. The preset duration can be a preset duration such as 20 seconds or 1 minute; after determining the abnormal time point, the preset duration can be extended before and after the abnormal time point to obtain a time period, such as 9:25:30 at this time, and extended 20 seconds before and after to obtain a time period from 9:25:10 to 9:25:50; or only based on the abnormal time point, the preset duration is extended forward or backward; or the preset duration is used as the length of the time period, and the abnormal time point is used as the midpoint of the preset duration to obtain a time period. The vehicle operation data and sound monitoring data within this time period are used as abnormal monitoring data. The vehicle side further locates the abnormal monitoring data, determines the time point when the abnormality occurs and the monitoring data of the vehicle when the abnormal noise occurs, so as to facilitate the subsequent troubleshooting and analysis of the abnormal noise fault and improve the rate of fault resolution.

[0058] S4: Uploading the abnormal monitoring data to the cloud, wherein the cloud is used to verify whether the vehicle has an abnormality according to the abnormal monitoring data.

[0059] Specifically, the cloud can collect audio data uploaded by staff and can also accept audio data uploaded by multiple vehicles. Based on the obtained data, the cloud-based deep autoencoder is regularly trained and updated. Therefore, the cloud-based deep autoencoder will be saved. The cloud-based abnormal monitoring data uploaded by the vehicle is input into the latest deep autoencoder to verify whether there is any abnormality in the vehicle. The verification results can be sent to the user or vehicle to increase the accuracy of diagnosis.

[0060] In this embodiment, the audio monitoring data of the vehicle is processed to obtain sound features, and then the sound features are input into the edge deep autoencoder to determine whether the vehicle is abnormal. The abnormal sound of the vehicle can be monitored locally on the vehicle based on the edge deep autoencoder, and the data does not need to be uploaded to the cloud. The data will not be delayed due to cloud transmission problems, and the local vehicle can respond to the abnormality in a timely manner to ensure the stability of monitoring; if it is determined that the vehicle is suspected of having an abnormality, the abnormal monitoring data is determined according to the time point of the abnormality, the preset duration and the operation data of the vehicle, and the abnormal monitoring data is further located to determine the time point of the abnormality and the monitoring data of the vehicle when the abnormal sound occurs, so as to facilitate the subsequent troubleshooting and analysis of the abnormal sound fault and improve the rate of fault resolution; the abnormal monitoring data is uploaded to the cloud, wherein the cloud verifies whether the vehicle is abnormal based on the abnormal monitoring data, for example, by using the latest deep autoencoder stored in the cloud for verification. After multiple verifications, the accuracy of monitoring is improved, and the vehicle and the cloud achieve a cloud-edge synergy effect, taking into account the timeliness, stability and accuracy of monitoring.

[0061] Alternatively, if Figure 2 As shown, the inputting of the sound feature into the edge deep autoencoder to determine whether the vehicle is abnormal includes:

[0062] The sound features are input into an edge deep autoencoder to obtain sound reconstruction features, wherein the edge deep autoencoder is obtained by the cloud side sending a deep autoencoder trained with a training set to the vehicle side, and the training set includes normal audio monitoring data.

[0063] Specifically, for example, a 1s audio monitoring data has a sampling frequency of 22050Hz, that is, the audio monitoring data is (22050, 1). After processing, the obtained sound feature may be (16, 44, 1). After reconstruction by the edge depth autoencoder, the obtained sound reconstruction feature may be (15.5, 44.5, 1). The purpose of sound feature reconstruction is to extract more representative and distinguishable features from the original sound signal in order to improve classification and recognition accuracy. The reconstructed features are usually more compact than the original sound data, so the amount of data can be reduced and efficiency can be improved during storage and transmission. Common methods for sound feature reconstruction include time-frequency analysis methods (such as short-time Fourier transform, Mel-frequency cepstral coefficients, etc.), filter bank-based methods (such as Mel-frequency cepstral coefficients) or autoencoders.

[0064] A loss value is obtained according to the sound reconstruction feature and the sound feature.

[0065] Specifically, since the edge deep autoencoder is trained on the cloud using normal audio monitoring data, the edge deep autoencoder has a better effect in reconstructing the sound features of normal audio monitoring data, but has a poorer effect in reconstructing the sound features of abnormal sound audio data. When the edge deep autoencoder reconstructs the sound features of normal audio monitoring data, the obtained sound reconstruction features are not much different from the input sound features; however, after the sound features of abnormal sound audio data are reconstructed, the gap between the sound features and the sound reconstruction features is relatively large, that is, the loss value is relatively large. In addition, the three-dimensional array of the sound reconstruction features is subtracted from the three-dimensional array of the sound features, and the sound difference is obtained after subtraction. The square of the sound difference is calculated to obtain the square difference, and the values ​​of the three dimensions of the square difference are added and averaged to obtain the mean square error, which is used as the loss value.

[0066] It is determined whether the loss value is greater than a reconstruction error threshold.

[0067] If the loss value is greater than the reconstruction error threshold, the vehicle is judged to be suspected of being abnormal, and a reminder message is sent to the user terminal. If the loss value is less than or equal to the reconstruction error threshold, the vehicle is judged to be normal.

[0068] Specifically, when the loss value is greater than the reconstruction error threshold, it means that the loss value is large, which can reversely confirm the moment corresponding to the audio monitoring data at this time. The vehicle is suspected to have an abnormality, and a reminder message can be sent to the user end to remind the user to pay attention to the vehicle's operating status in time, or remind the user to listen for abnormal sounds, slow down or stop for inspection, etc., to avoid safety hazards when continuing to drive.

[0069] Alternatively, if Figure 2As shown, if it is determined that the vehicle is suspected of having an abnormality, the abnormal monitoring data is determined according to the time point when the vehicle has an abnormality, the preset duration and the operation data of the vehicle, including:

[0070] If it is determined that the vehicle is suspected of having an abnormality, the abnormal time period is determined based on the time point of the abnormality and the preset duration.

[0071] Specifically, the preset duration can be a pre-set duration such as 20 seconds or 1 minute; after determining the abnormal time point, the preset duration can be extended before and after the abnormal time point to obtain a time period, for example, at this time, 9:25:30, extended before and after 20 seconds, to obtain a time period from 9:25:10 to 9:25:50; or only based on the abnormal time point, the preset duration is extended forward or backward to obtain a time period from 9:25:10 to 9:25:30 or a time period from 9:25:30 to 9:25:50; or the preset duration is used as the length of the time period, and the abnormal time point is used as the midpoint of the preset duration, to obtain a time period from 9:25:20 to 9:25:40.

[0072] The audio monitoring data and the vehicle operation data corresponding to the abnormal time period are marked as suspected abnormal audio monitoring data and suspected abnormal vehicle operation data, respectively.

[0073] Specifically, after determining the abnormal time period, the audio monitoring data and vehicle operation data corresponding to the abnormal time period are marked as suspected abnormal audio monitoring data and suspected abnormal vehicle operation data, respectively. They are temporarily marked as suspected abnormal here because the vehicle driving environment is complex. Sometimes the vehicle may pass through noisy roads, or pass near construction sites, etc. At this time, the sound collection equipment will record the ambient noise together. To avoid strong interference from ambient noise, the sound can be preprocessed and cleaned, but the data preprocessing may not be in place or the preprocessing may fail to completely remove the ambient noise. When abnormal sound monitoring and judgment are performed based on these audio monitoring data containing noisy ambient sounds, the results may not necessarily be accurate. At this time, it is necessary to temporarily mark the abnormality judged only based on the audio monitoring data as suspected abnormality.

[0074] Determine whether the suspected abnormal vehicle operation data meets the preset abnormal rules.

[0075] Specifically, the vehicle's operating data may include the operating data of the motor, the engine, the transmission, and the wiper, etc. The operating data of these components can be compared with the data of normal operation. For example, when the motor is at a certain voltage value, there is a corresponding normal operating temperature and normal speed. However, the operating temperature or speed of the motor is now much different from the values ​​under normal conditions. At this time, it can be determined that the motor is abnormal. At this time, it can at least be determined that the motor may be faulty. If the abnormality is detected at this time, it is highly likely that the abnormal sound is indeed emitted by the vehicle itself, rather than caused by noise such as environmental noise.

[0076] If the suspected abnormal vehicle operation data meets the preset abnormal rules, the abnormal monitoring data is determined, wherein the abnormal monitoring data includes the suspected abnormal audio monitoring data and the suspected abnormal vehicle operation data. When the suspected abnormal vehicle operation data does not meet the preset abnormal rules, it is determined that the vehicle is normal, there is no abnormal monitoring data, and the data is not uploaded.

[0077] In this embodiment, the edge deep autoencoder reconstructs the sound features. After performing abnormal diagnosis based on the sound reconstruction features and the sound features, a secondary judgment is further performed in combination with the vehicle operation data to avoid the problem of inaccurate judgment results caused by noise in the audio monitoring data. After the double diagnosis, the determined abnormal monitoring data is more accurate, reducing the upload of invalid data to the cloud, avoiding frequent occupation of data transmission channels, ensuring the validity and accuracy of the uploaded data, and saving cloud processing capabilities.

[0078] Alternatively, if Figure 2 As shown, the suspected abnormal vehicle operation data includes fault codes and operation control parameters.

[0079] The determining whether the suspected abnormal vehicle operation data satisfies a preset abnormality rule includes:

[0080] Inputting the operation control parameters corresponding to the various components of the vehicle into the corresponding component diagnosis model to diagnose whether each of the components is operating abnormally;

[0081] Specifically, the component diagnosis model is a binary classification model sent down after training on the cloud. The cloud uses the vehicle parameter data collected offline in the vehicle laboratory under all working conditions to train a binary classification model, and the training results are normal operation and abnormal operation. The trained model is sent to the vehicle end as a component diagnosis model to diagnose and identify the operation control parameters of the component. It is also possible to directly set the component diagnosis model based on experience or directly migrate mature component diagnosis models from previous vehicles, and then use the component diagnosis model to monitor and judge the operation control parameters of the component, identify whether the operation control parameters of the component are abnormal, and expand the coverage of the diagnosis.

[0082] The total number of abnormalities is obtained according to the number of the fault codes in the suspected abnormal vehicle operation data and the number of the components diagnosed with operating abnormalities.

[0083] Specifically, the vehicle is equipped with monitoring equipment for each component, or each component is equipped with a self-monitoring module, which can monitor common faults that may occur in the component. When a component is monitored to have a fault, a fault code will be generated. By querying the fault code corresponding table, the specific fault of the component can be known. For example, if the servo motor has a fault code 10, it means that the position error of the servo motor exceeds the limit. If it continues to run, the position control accuracy of the servo motor is not good. Therefore, the fault code in the vehicle operation data can also be used to infer that a component in the vehicle has a fault. By counting the number of fault codes and the number of components with abnormal operation diagnosed by the component diagnosis model, and adding the two numbers, the number of fault conditions, that is, the total number of abnormalities, can be known.

[0084] Determine whether the total number of anomalies is greater than a preset number threshold.

[0085] If the total number of abnormalities is greater than the preset number threshold, the suspected abnormal vehicle operation data meets the preset abnormal rule, and the suspected abnormal audio monitoring data and the suspected abnormal vehicle operation data are determined as the abnormal monitoring data. When the total number of abnormalities is less than or equal to the preset number threshold, the vehicle is determined to be normal and no data is uploaded.

[0086] Specifically, in order to avoid missed faults, the preset quantity threshold can be set smaller, such as 1 or 2; in addition, in the initial stage of operation on the vehicle side or cloud side, the preset quantity threshold can be set smaller, such as 0. As long as a fault code appears or a component abnormality is diagnosed, the abnormal monitoring data will be uploaded, so that the cloud can obtain more data samples, which is conducive to data analysis on the cloud side, or training and updating the binary classification model so that the binary classification model can identify various fault conditions, and the updated binary classification model will be sent to the vehicle side as a component diagnosis model to make the diagnosis of the component more accurate.

[0087] Alternatively, if Figure 2 As shown, before uploading the abnormal monitoring data to the cloud, the vehicle abnormal noise monitoring data processing method further includes:

[0088] Determine whether user authorization information is received.

[0089] When the user authorization information is received and the abnormal monitoring data is desensitized, the abnormal monitoring data is uploaded to the cloud.

[0090] In this embodiment, in order to protect the privacy of the user, the abnormal monitoring data cannot be uploaded to the cloud without the user's authorization, even if an abnormal noise is detected in the vehicle. Only after receiving the user's authorization instruction, the abnormal monitoring data can be uploaded to the cloud when the vehicle has an abnormality. However, the abnormal monitoring data also needs to be desensitized before uploading, such as removing the human voice in the abnormal audio monitoring data, or blurring the human voice, etc., to avoid leakage of the conversation content of users in the car.

[0091] Optionally, before inputting the sound feature into the edge deep autoencoder to determine whether the vehicle is abnormal, the vehicle abnormal sound monitoring data processing method further includes:

[0092] According to the deep autoencoder periodically sent down by the cloud, the edge deep autoencoder is updated, wherein the cloud is used to extract the sound features of the normal audio monitoring data in the training set, input the sound features into the deep autoencoder, obtain the reconstructed sound features, and compare the sound features with the reconstructed sound features; if the loss value between the reconstructed sound features and the sound features is greater than or equal to the allowable loss value, adjust the parameters of the deep autoencoder and continue training; if the loss value between the reconstructed sound features and the sound features is less than the allowable loss value, stop training, and obtain the trained deep autoencoder, and the training set includes the normal audio monitoring data obtained from the vehicle end.

[0093] Specifically, Figure 3As shown, the deep autoencoder can be an LSTM deep autoencoder, and the LSTM deep autoencoder includes an LSTM encoder, a hidden layer, and an LSTM reconstructor; the LSTM encoder, the hidden layer, and the LSTM reconstructor are connected in this way. The training set is formed by using the normal audio monitoring data in the car collected from the background or from multiple user car terminals, and the training set is continuously updated or supplemented. The deep autoencoder is continuously trained using the training set until the sound error (i.e., the loss value) between the sound reconstruction feature and the input sound feature is within an acceptable loss range. At this time, the training can be stopped, and the updated deep autoencoder is obtained after training. The updated deep autoencoder is sent to the car terminal. As the edge deep autoencoder of the car terminal, the deep autoencoder can be retrained and updated regularly, or when the number of data samples in the training set reaches the required number, the deep autoencoder can be automatically retrained and updated.

[0094] Optionally, before inputting the sound feature into the edge deep autoencoder to determine whether the vehicle is abnormal, the vehicle abnormal sound monitoring data processing method further includes:

[0095] The reconstruction error threshold of the vehicle is updated according to the reconstruction error threshold periodically sent down by the cloud, wherein the reconstruction error threshold periodically sent down by the cloud is greater than a preset proportion of the loss values ​​in a loss value set, and the loss value set is obtained by the loss value of the deep autoencoder corresponding to each normal audio monitoring data in the cloud statistical test set.

[0096] Specifically, after the deep autoencoder is trained in the cloud, it is tested using a test set. The loss value corresponding to each normal audio monitoring data in the test set is counted, and the distribution of all loss values ​​is plotted into a histogram. The horizontal axis of the histogram is the loss value, and the vertical axis is the probability density. The reconstruction error threshold is selected based on the condition that it can cover more than 95% of the loss values, that is, the reconstruction error threshold is greater than more than 95% of the statistical loss values.

[0097] like Figure 4 As shown, a vehicle end 100 provided by an embodiment of the present invention includes:

[0098] The data processing module 110 is used to process the audio monitoring data of the vehicle to obtain sound features.

[0099] Specifically, at the data source level, sound signals (audio monitoring data), vehicle controller signals (operation control parameters of various components of the vehicle) and vehicle diagnostic fault signals (fault codes) are collected. The external sound collection device is connected to the corresponding signal collection device or signal collection module through the reserved interface of the vehicle end (such as the sound collection smart box). The data processing module 110 performs corresponding processing according to the data type, wherein the data processing module 110 may also include a sound processing module, a controller signal processing module and a fault code processing module; the sound processing module processes and extracts the sound features of the audio monitoring data, the controller signal processing module performs data parsing, obtains the vehicle operation control parameters, and the fault code processing module extracts the fault code and queries the mapping content of the fault code. In addition, public operations can also be performed on each type of data, and the public operations are data timestamp marking or public field completion, such as adding vehicle model information. The data processing module 110 sends the processed sound features to the edge diagnosis module 120.

[0100] The edge diagnosis module 120 is used to input the sound features into the edge deep autoencoder to determine whether the vehicle has an abnormality, wherein the edge deep autoencoder is arranged on the vehicle side; if it is determined that the vehicle is suspected of having an abnormality, the abnormal monitoring data is determined according to the time point when the vehicle has an abnormality, a preset duration and the operating data of the vehicle.

[0101] Specifically, after receiving the sound features, the edge diagnosis module 120 can first perform data start time alignment and intermediate time interpolation to facilitate the alignment of the sound features at the same time with the vehicle's operating data. The deep autoencoder sent from the cloud is used as the edge deep autoencoder to perform abnormality detection on the vehicle side. The cloud will periodically upgrade the edge deep autoencoder on the vehicle side. If the vehicle side detects an abnormality in the vehicle, the abnormal result will be fed back to the client in real time via Bluetooth.

[0102] The main control module 130 is used to upload the abnormal monitoring data to the cloud, wherein the cloud is used to verify whether the vehicle has an abnormality based on the abnormal monitoring data.

[0103] Specifically, data will not be automatically uploaded to the cloud before user authorization; after obtaining user authorization and allowing information to be uploaded to the cloud, the main control module 130 receives the request and will require the edge diagnosis module 120 to start desensitizing the abnormal monitoring data in real time and push the abnormal monitoring data to the message queue in the cloud. The uploaded abnormal monitoring data may include real-time diagnostic results, sound characteristics, fault codes, etc.

[0104] In this embodiment, the data processing module 110 processes the audio monitoring data of the vehicle to obtain sound features, and then the edge diagnosis module 120 inputs the sound features into the edge deep autoencoder to determine whether the vehicle is abnormal. The vehicle abnormal sound can be monitored locally on the vehicle side based on the edge deep autoencoder, without uploading the data to the cloud. The data will not be delayed due to cloud transmission problems. The local vehicle side can respond to abnormalities in a timely manner to ensure the stability of monitoring, support users to diagnose and continuously optimize the encoder even when they are not connected to the Internet, and better match users' expectations of privacy protection. If it is determined that the vehicle is suspected of being abnormal, according to the The time point when the vehicle abnormality occurs, the preset duration and the operating data of the vehicle are used to determine the abnormal monitoring data, and further locate the abnormal monitoring data to determine the time point when the abnormality occurs and the monitoring data of the vehicle when the abnormal sound occurs, so as to facilitate the subsequent troubleshooting and analysis of the abnormal sound fault and improve the rate of fault resolution; the main control module 130 uploads the abnormal monitoring data to the cloud, wherein the cloud verifies whether the vehicle has an abnormality based on the abnormal monitoring data, for example, using the latest deep autoencoder stored in the cloud for verification. After multiple verifications, the accuracy of monitoring is improved, and the vehicle-side and cloud-edge synergy effect is achieved, taking into account the timeliness, stability and accuracy of monitoring.

[0105] Optionally, the vehicle side may also include a data acquisition module 140 and a data storage module 150. The data acquisition module 140 receives a hexadecimal signal transmitted from an external sound collection device and stores the original data in the data storage module 150. The data storage module 150 is divided into two parts, namely, a memory storage queue and a disk space. After the data is sent to the memory storage queue, the data is automatically pushed to the corresponding data processing module 110. At the same time, the data is stored in the disk space, and the data will be processed to disk to respond to the original data request from the cloud. In addition, the data storage is periodic overwriting storage, because the amount of data after parsing is 2 to 3 times that of the original data, and the original data is stored instead of the processed data, which reduces the storage pressure on the vehicle side.

[0106] Optionally, a vehicle abnormal noise monitoring data processing method provided by an embodiment of the present invention is applied to the cloud, including:

[0107] According to the abnormal monitoring data sent by the vehicle side, check whether the vehicle has any abnormality, wherein the abnormal monitoring data is processed by the vehicle side on the audio monitoring data of the vehicle to obtain sound features, and the sound features are input into the edge deep autoencoder to determine whether the vehicle has any abnormality. If it is determined that the vehicle is suspected of having an abnormality, it is determined based on the time point when the vehicle has the abnormality, the preset duration and the operating data of the vehicle.

[0108] like Figure 4As shown, a cloud 200 provided by an embodiment of the present invention includes:

[0109] The data verification module 210 is used to verify whether the vehicle has an abnormality based on the abnormal monitoring data sent by the vehicle end, wherein the vehicle end processes the audio monitoring data of the vehicle to obtain sound features; the sound features are input into the edge deep autoencoder to determine whether the vehicle has an abnormality; if it is determined that the vehicle is suspected of having an abnormality, the abnormal monitoring data is determined based on the time point when the vehicle has an abnormality, a preset duration and the operating data of the vehicle.

[0110] Optionally, the cloud also includes a message queue module 220 and a training module 230. The message queue module 220 is used to receive the abnormal monitoring data uploaded by the vehicle side, and push the data to the data verification module 210 in real time. The data verification module 210 cleans and analyzes the data, and diagnoses according to the latest updated deep autoencoder of the training module 230. In addition, the vehicle side can also upload the abnormal diagnosis results first. After receiving the abnormal diagnosis results in the cloud, it will notify the main control module 130 of the vehicle side through RPC (Remote Procedure Call, remote procedure call), requesting to upload all the abnormal monitoring data to the cloud. After receiving the request message, the main control module 130 forwards the request message to the data storage module 150. The data storage module 150 will upload the original monitoring data to the message queue module 220, specify TOPIC (a high-level classification of the message, here for data interaction), and push it to the data verification module 210. After cleaning and parsing, the data verification module 210 performs cloud-edge collaborative diagnosis to increase the accuracy of diagnosis. At the same time, the data is input into the training module 230 to participate in the training of the binary classification model.

[0111] The training module 230 extracts the sound features of the normal audio monitoring data in the training set, inputs the sound features into the deep autoencoder, obtains the reconstructed sound features, and compares the sound features with the reconstructed sound features; if the loss value between the reconstructed sound features and the sound features is greater than or equal to the allowable loss value, adjusts the parameters of the deep autoencoder and continues training; if the loss value between the reconstructed sound features and the sound features is less than the allowable loss value, stops training, and obtains the trained deep autoencoder. The training module 230 is also used to count the loss values ​​of the deep autoencoder corresponding to each normal audio monitoring data in the test set to form a loss value set, and the reconstruction error threshold is greater than the loss value of a preset proportion in the loss value set. After training and optimization, the training module 230 will update the deep autoencoder to the data verification module 210, and send it to the vehicle end for edge deep autoencoder upgrade.

[0112] In this embodiment, the vehicle-side 100 and the cloud-side 200 interact and collaborate. Without user authorization, the vehicle-side 100 diagnosis serves as effective support to protect user privacy. Even without authorization, the cloud-side 200 will send the updated deep autoencoder to the vehicle-side 100 in a unidirectional manner to ensure that the edge deep autoencoder is continuously optimized to improve the accuracy of diagnosis. With user authorization, the vehicle-side routinely only uploads the diagnosis results. When the cloud-side 200 identifies that the vehicle-side 100 has a suspected abnormality, it sends a request message to the vehicle-side 100 to upload the abnormal monitoring information of the vehicle-side 100 (i.e., the data before and after the abnormal situation in N seconds) to improve data transmission efficiency and reduce data loss rate. The diagnostic stability is supported by local diagnosis on the vehicle-side 100, and the cloud-side 200 continuously optimizes the deep autoencoder. The cloud-edge collaboration ensures the effectiveness of the diagnosis.

[0113] In addition, the vehicle end 100 adopts a centralized acquisition method, and the vehicle end 100 and the cloud end 200 perform clock alignment. The vehicle end 100 is aligned with the same clock, and the acquisition time of the sound signal and the vehicle parameters is highly consistent, and the error is reduced to the millisecond level. In addition to using the CAN communication protocol, the external acquisition device and the vehicle end can also add XCP (Universal Measurement and Calibration Protocol), UDS (Unified Diagnostic Services), OBD (On-Board Diagnostics) transmission protocols to increase data diversity, expand the breadth of data sources, further improve the efficiency of sound and vehicle parameter acquisition, and broaden the range of tested vehicle types, supporting mass-produced vehicles and test vehicles. The vehicle basic signal acquisition adopts the XCP protocol, including XCP-ON-CAN, XCP-ON-ETH (Ethernet), etc. The CAN communication protocol will transmit about 20% of parameters that are not directly related to diagnosis, avoiding the waste of transmission channels caused by the CAN communication protocol transmission, thereby increasing the data utilization rate by about 20%.

[0114] like Figure 5 As shown, a vehicle 500 provided by an embodiment of the present invention includes a memory 510 and a processor 520; the memory 510 is used to store a computer program; the processor 520 is used to implement the vehicle abnormal sound monitoring data processing method as described above when executing the computer program.

[0115] In other words, a vehicle 500 includes a memory 510 and a processor 520 coupled to the memory 510; the memory 510 is configured to store a computer program; and the processor 520 is configured to perform the following operations when executing the computer program:

[0116] Processing the audio monitoring data of the vehicle to obtain sound characteristics;

[0117] Inputting the sound features into an edge deep autoencoder to determine whether the vehicle is abnormal;

[0118] If it is determined that the vehicle is suspected of having an abnormality, determining abnormal monitoring data according to the time point when the vehicle has an abnormality, a preset duration, and the operating data of the vehicle;

[0119] The abnormal monitoring data is uploaded to the cloud, wherein the cloud verifies whether the vehicle has an abnormality according to the abnormal monitoring data.

[0120] Alternatively, based on the abnormal monitoring data sent by the vehicle side, check whether the vehicle has an abnormality, wherein the abnormal monitoring data is processed by the vehicle side on the audio monitoring data of the vehicle to obtain sound features, and the sound features are input into the edge deep autoencoder to determine whether the vehicle has an abnormality. If it is determined that the vehicle is suspected of having an abnormality, it is determined based on the time point when the vehicle has an abnormality, the preset duration and the operating data of the vehicle.

[0121] The vehicle 500 may be equipped with various forms of digital electronic computer devices, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The examples described above may be used as electronic devices on the vehicle 500 of the present invention, which are examples of hardware devices that may be applied to various aspects of the present invention. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein.

[0122] The vehicle 500 includes a computing unit that can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) or a computer program loaded from a storage unit into a random access memory (RAM). In the RAM, various programs and data required for the operation of the device can also be stored. The computing unit, ROM, and RAM are connected to each other via a bus. An input / output (I / O) interface is also connected to the bus.

[0123] An embodiment of the present invention provides a computer-readable storage medium, wherein a computer program is stored on the storage medium. When the computer program is executed by a processor, the following operations are performed:

[0124] Processing the audio monitoring data of the vehicle to obtain sound characteristics;

[0125] Inputting the sound features into an edge deep autoencoder to determine whether the vehicle is abnormal;

[0126] If it is determined that the vehicle is suspected of having an abnormality, determining abnormal monitoring data according to the time point when the vehicle has an abnormality, a preset duration, and the operating data of the vehicle;

[0127] The abnormal monitoring data is uploaded to the cloud, wherein the cloud verifies whether the vehicle has an abnormality according to the abnormal monitoring data.

[0128] Alternatively, based on the abnormal monitoring data sent by the vehicle side, check whether the vehicle has an abnormality, wherein the abnormal monitoring data is processed by the vehicle side on the audio monitoring data of the vehicle to obtain sound features, and the sound features are input into the edge deep autoencoder to determine whether the vehicle has an abnormality. If it is determined that the vehicle is suspected of having an abnormality, it is determined based on the time point when the vehicle has an abnormality, the preset duration and the operating data of the vehicle.

[0129] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM) or a random access memory (RAM), etc. In the present application, the unit described as a separate component may or may not be physically separated, and the component displayed as a unit may or may not be a physical unit, that is, it may be located in one place, or it may be distributed on multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiment of the present invention. In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0130] Although the present invention is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the scope defined by the claims.

Claims

1. A vehicle abnormal noise monitoring data processing method, characterized in that: Applied to the vehicle side, including: Processing the audio monitoring data of the vehicle to obtain sound characteristics; Inputting the sound feature into an edge depth autoencoder to determine whether the vehicle is abnormal, wherein the edge depth autoencoder is arranged at the vehicle end; If it is determined that the vehicle is suspected of having an abnormality, determining abnormal monitoring data according to the time point when the vehicle has an abnormality, a preset duration, and the operating data of the vehicle; The abnormal monitoring data is uploaded to the cloud, wherein the cloud is used to verify whether the vehicle has an abnormality based on the abnormal monitoring data.

2. The vehicle abnormal noise monitoring data processing method according to claim 1, characterized in that: Inputting the sound feature into the edge deep autoencoder to determine whether the vehicle is abnormal includes: Inputting the sound feature into an edge deep autoencoder to obtain a sound reconstruction feature, wherein the edge deep autoencoder is obtained by the cloud sending a deep autoencoder trained with a training set to the vehicle end, wherein the training set includes normal audio monitoring data; Obtaining a loss value according to the sound reconstruction feature and the sound feature; Determining whether the loss value is greater than a reconstruction error threshold; If the loss value is greater than the reconstruction error threshold, it is determined that the vehicle is suspected of being abnormal.

3. The vehicle abnormal noise monitoring data processing method according to claim 1, characterized in that: If it is determined that the vehicle is suspected of having an abnormality, determining the abnormal monitoring data according to the time point when the vehicle has an abnormality, a preset duration, and the operation data of the vehicle includes: If it is determined that the vehicle is suspected of having an abnormality, determining an abnormal time period according to the abnormal time point and the preset duration; Marking the audio monitoring data and the vehicle operation data corresponding to the abnormal time period as suspected abnormal audio monitoring data and suspected abnormal vehicle operation data, respectively; Determining whether the suspected abnormal vehicle operation data meets a preset abnormality rule; If the suspected abnormal vehicle operation data satisfies the preset abnormal rule, the abnormal monitoring data is determined, wherein the abnormal monitoring data includes the suspected abnormal audio monitoring data and the suspected abnormal vehicle operation data.

4. The vehicle abnormal noise monitoring data processing method according to claim 3 is characterized in that: The suspected abnormal vehicle operation data includes fault codes and operation control parameters; The determining whether the suspected abnormal vehicle operation data satisfies a preset abnormality rule includes: Inputting the operation control parameters corresponding to the various components of the vehicle into the corresponding component diagnosis model to diagnose whether each of the components is operating abnormally; Obtaining a total number of abnormalities according to the number of the fault codes in the suspected abnormal vehicle operation data and the number of the components diagnosed with abnormal operation; Determine whether the total number of abnormalities is greater than a preset number threshold; If the total number of anomalies is greater than the preset number threshold, the suspected abnormal vehicle operation data satisfies the preset abnormality rule.

5. The vehicle abnormal noise monitoring data processing method according to claim 1, characterized in that: Before inputting the sound feature into the edge deep autoencoder to determine whether the vehicle is abnormal, the method further includes: According to the deep autoencoder periodically sent down by the cloud, the edge deep autoencoder is updated, wherein the cloud is used to extract the sound features of the normal audio monitoring data in the training set, input the sound features into the deep autoencoder, obtain the reconstructed sound features, and compare the sound features with the reconstructed sound features; if the loss value between the reconstructed sound features and the sound features is greater than or equal to the allowable loss value, adjust the parameters of the deep autoencoder and continue training; if the loss value between the reconstructed sound features and the sound features is less than the allowable loss value, stop training, and obtain the trained deep autoencoder, and the training set includes the normal audio monitoring data obtained from the vehicle end.

6. The vehicle abnormal noise monitoring data processing method according to claim 2, characterized in that: Before inputting the sound feature into the edge deep autoencoder to determine whether the vehicle is abnormal, the method further includes: The reconstruction error threshold of the vehicle is updated according to the reconstruction error threshold periodically sent down by the cloud, wherein the reconstruction error threshold periodically sent down by the cloud is greater than a preset proportion of the loss values ​​in a loss value set, and the loss value set is obtained by the loss value of the deep autoencoder corresponding to each normal audio monitoring data in the cloud statistical test set.

7. The vehicle abnormal noise monitoring data processing method according to claim 1, characterized in that: Before uploading the abnormal monitoring data to the cloud, the method further includes: Determine whether user authorization information is received; When the user authorization information is received and the abnormal monitoring data is desensitized, the abnormal monitoring data is uploaded to the cloud.

8. A vehicle end, characterized in that: include: A data processing module, used to process the audio monitoring data of the vehicle to obtain sound characteristics; an edge diagnosis module, configured to input the sound feature into an edge deep autoencoder to determine whether the vehicle is abnormal, wherein the edge deep autoencoder is arranged at the vehicle end; if it is determined that the vehicle is suspected of being abnormal, determining abnormal monitoring data according to the time point when the vehicle is abnormal, a preset duration and the operation data of the vehicle; The main control module is used to upload the abnormal monitoring data to the cloud, wherein the cloud is used to verify whether the vehicle has an abnormality based on the abnormal monitoring data.

9. A vehicle abnormal noise monitoring data processing method, characterized in that: Applied in the cloud, including: According to the abnormal monitoring data sent by the vehicle side, check whether the vehicle has any abnormality, wherein the abnormal monitoring data is processed by the vehicle side on the audio monitoring data of the vehicle to obtain sound features, and the sound features are input into the edge deep autoencoder to determine whether the vehicle has any abnormality. If it is determined that the vehicle is suspected of having an abnormality, it is determined based on the time point when the vehicle has the abnormality, the preset duration and the operating data of the vehicle.

10. A cloud, characterized in that: include: The data verification module is used to verify whether the vehicle has any abnormality based on the abnormal monitoring data sent by the vehicle side, wherein the abnormal monitoring data is processed by the vehicle side on the audio monitoring data of the vehicle to obtain sound features, and the sound features are input into the edge deep autoencoder to determine whether the vehicle has any abnormality. If it is determined that the vehicle is suspected of having an abnormality, it is determined based on the time point when the vehicle has the abnormality, the preset duration and the operating data of the vehicle.

11. A vehicle, characterized in that: It includes a memory and a processor, the memory is used to store a computer program, and the processor is used to implement the vehicle abnormal sound monitoring data processing method as described in any one of claims 1 to 7 or the vehicle abnormal sound monitoring data processing method as described in claim 9 when executing the computer program.

Images