Automobile fault monitoring and diagnosing system and method based on internet of things

Abstract

The application provides an Internet of Things-based automobile fault monitoring and diagnosis system and method, and relates to the field of intelligent automobile fault monitoring. First, real-time operation data of various components of an automobile is collected through various sensors. Then, the collected operation data of various components of the automobile is analyzed in real time by using an artificial intelligence-based data analysis technology, so as to determine whether the automobile has hidden faults. In this way, more accurate and efficient automobile fault monitoring can be realized, thereby providing strong guarantee for the safe operation of new energy automobiles.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent automobile fault monitoring, and more specifically, to an automobile fault monitoring and diagnosis system and method based on the Internet of Things. BACKGROUND

[0002] New energy vehicles refer to vehicles that use unconventional vehicle fuels as power sources and integrate advanced technologies in vehicle power control and driving to form vehicles with advanced technical principles and new technologies and structures. Existing new energy vehicles are inspected before leaving the factory, and after being used by customers for a period of time, they need to be maintained and diagnosed by service providers as required to ensure normal and safe operation. At present, new energy vehicle maintenance institutions often use diagnostic instruments for vehicle fault diagnosis, and a large amount of time is required for detecting each component of the new energy vehicle, the diagnosis cost is high, and at present, fault monitoring is not completely subdivided into parameter diagnosis of each system such as electric drive, battery, and braking, so there is a certain probability of misdiagnosis in diagnosis, resulting in the replacement of wrong parts.

[0003] Therefore, an optimized automobile fault monitoring and diagnosis scheme is needed. SUMMARY

[0004] To solve the above technical problems, the present application is proposed. The embodiments of the present application provide an automobile fault monitoring and diagnosis system and method based on the Internet of Things, which first collects real-time operation data of each component of the automobile through various sensors, then uses artificial intelligence-based data analysis technology to analyze the collected operation data of each component of the automobile in real time, and further judges whether the automobile has hidden faults. In this way, more accurate and efficient automobile fault monitoring can be achieved, thereby providing strong protection for the safe operation of new energy vehicles.

[0005] According to one aspect of the present application, an automobile fault monitoring and diagnosis system based on the Internet of Things is provided, which comprises:

[0006] An automobile fault monitoring related data collection module is configured to obtain driving motor speed values, driving motor vibration frequency values, battery pack voltage values, battery pack current values, battery pack internal resistance values, battery pack temperature values, four sets of tire pressure values, brake pad temperature values, brake pad thickness values, and brake pad sound frequency values at a plurality of predetermined time points within a predetermined time period.

[0007] An automobile electric power drive monitoring module is configured to time sequence encode the driving motor speed values and the driving motor vibration frequency values at the plurality of predetermined time points to obtain an automobile electric power drive monitoring time sequence feature matrix.

[0008] An automobile battery monitoring module is configured to time-series encode the battery pack voltage values, battery pack current values, battery pack internal resistance values, and battery pack temperature values at the plurality of predetermined time points to obtain an automobile battery monitoring time-series feature matrix.

[0009] An automobile tire monitoring module is configured to time-series encode the four sets of tire pressure values at the plurality of predetermined time points to obtain an automobile tire pressure monitoring time-series feature matrix.

[0010] An automobile brake pad monitoring module is configured to time-series encode the brake pad temperature values, brake pad thickness values, and brake pad sound frequency values at the plurality of predetermined time points to obtain an automobile brake pad monitoring time-series feature matrix.

[0011] An automobile monitoring feature fusion module is configured to fuse the automobile power drive monitoring time-series feature matrix, the automobile battery monitoring time-series feature matrix, the automobile tire pressure monitoring time-series feature matrix, and the automobile brake pad monitoring time-series feature matrix to obtain an automobile monitoring feature matrix.

[0012] An automobile fault monitoring result generation module is configured to obtain an automobile fault monitoring result based on the automobile monitoring feature matrix.

[0013] According to another aspect of the present application, an automobile fault monitoring and diagnosis method based on the Internet of Things is provided, which includes:

[0014] Obtaining drive motor rotation speed values, drive motor vibration frequency values, battery pack voltage values, battery pack current values, battery pack internal resistance values, battery pack temperature values, four sets of tire pressure values, brake pad temperature values, brake pad thickness values, and brake pad sound frequency values at a plurality of predetermined time points within a predetermined time period.

[0015] Time-series encoding the drive motor rotation speed values and the drive motor vibration frequency values at the plurality of predetermined time points to obtain an automobile power drive monitoring time-series feature matrix.

[0016] Time-series encoding the battery pack voltage values, battery pack current values, battery pack internal resistance values, and battery pack temperature values at the plurality of predetermined time points to obtain an automobile battery monitoring time-series feature matrix.

[0017] Time-series encoding the four sets of tire pressure values at the plurality of predetermined time points to obtain an automobile tire pressure monitoring time-series feature matrix.

[0018] Time-series encoding the brake pad temperature values, brake pad thickness values, and brake pad sound frequency values at the plurality of predetermined time points to obtain an automobile brake pad monitoring time-series feature matrix.

[0019] fuse the automobile power drive monitoring time sequence feature matrix, the automobile battery monitoring time sequence feature matrix, the automobile tire pressure monitoring time sequence feature matrix and the automobile brake pad monitoring time sequence feature matrix to obtain an automobile monitoring feature matrix;

[0020] obtain an automobile fault monitoring result based on the automobile monitoring feature matrix.

[0021] Compared with the prior art, the automobile fault monitoring and diagnosis system and method based on the Internet of Things provided in the application can first collect real-time operation data of various components of the automobile through various sensors, then perform real-time analysis on the collected operation data of various components of the automobile by using the data analysis technology based on artificial intelligence, and further determine whether the automobile has hidden faults. In this way, more accurate and efficient automobile fault monitoring can be realized, thereby providing strong guarantee for safe operation of new energy automobiles. BRIEF DESCRIPTION OF DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor. In the drawings:

[0023] Figure 1 The system block diagram of the automobile fault monitoring and diagnosis system based on the Internet of Things according to the embodiment of the application.

[0024] Figure 2 The block diagram of the automobile power drive monitoring module in the automobile fault monitoring and diagnosis system based on the Internet of Things according to the embodiment of the application.

[0025] Figure 3 The block diagram of the automobile battery monitoring module in the automobile fault monitoring and diagnosis system based on the Internet of Things according to the embodiment of the application.

[0026] Figure 4 The block diagram of the automobile tire monitoring module in the automobile fault monitoring and diagnosis system based on the Internet of Things according to the embodiment of the application.

[0027] Figure 5 The block diagram of the automobile brake pad monitoring module in the automobile fault monitoring and diagnosis system based on the Internet of Things according to the embodiment of the application.

[0028] Figure 6 The block diagram of the automobile fault monitoring result generation module in the automobile fault monitoring and diagnosis system based on the Internet of Things according to the embodiment of the application.

[0029] Figure 7A flowchart of a method for monitoring and diagnosing automobile faults based on Internet of Things according to an embodiment of the present application. DETAILED DESCRIPTION

[0030] Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part but not all of the embodiments of the present application, and the present application can be implemented in many different ways. Therefore, the contents described herein should be considered as illustrative rather than limiting the present application.

[0031] New energy vehicles are powered by unconventional fuels, combining advanced technologies in vehicle power control and driving. They have the characteristics of advanced technical principles and novel structure. At present, new energy vehicles are strictly inspected before leaving the factory, and then need to be maintained and diagnosed at service providers according to requirements after being used by customers for a period of time, so as to ensure the normal and safe operation of the vehicles. At present, diagnostic instruments are often used in new energy vehicle maintenance agencies to diagnose vehicle faults. However, when detecting each component of the new energy vehicle, a large amount of time is often consumed, and the diagnosis cost is high. Moreover, the current fault monitoring is not completely refined to the parameter diagnosis level of each system such as electric drive, battery and braking. Therefore, there is a certain probability of misdiagnosis in the diagnosis process, which may lead to the phenomenon of replacing wrong parts. Therefore, an optimized automobile fault monitoring and diagnosis scheme is expected.

[0032] In recent years, deep learning and neural networks have been widely used in computer vision, natural language processing, text signal processing and other fields. In addition, deep learning and neural networks have shown a level close to or even beyond human in image classification, object detection, semantic segmentation, text translation and other fields. The development of deep learning and neural networks provides new solutions and schemes for automobile fault monitoring and diagnosis.

[0033] Figure 1 A system block diagram of a system for monitoring and diagnosing automobile faults based on Internet of Things according to an embodiment of the present application. As shown in Figure 1As shown, in the automobile fault monitoring and diagnosis system 100 based on the Internet of Things, it comprises: an automobile fault monitoring related data acquisition module 110, used for acquiring driving motor speed value, driving motor vibration frequency value, battery pack voltage value, battery pack current value, battery pack internal resistance value, battery pack temperature value, four sets of tire pressure value, brake pad temperature value, brake pad thickness value, brake pad sound frequency value at multiple predetermined time points within a predetermined time period; an automobile electric drive monitoring module 120, used for time sequence coding of the driving motor speed value and the driving motor vibration frequency value at the multiple predetermined time points to obtain an automobile electric drive monitoring time sequence feature matrix; an automobile battery monitoring module 130, used for time sequence coding of the battery pack voltage value, the battery pack current value, the battery pack internal resistance value and the battery pack temperature value at the multiple predetermined time points to obtain an automobile battery monitoring time sequence feature matrix; an automobile tire monitoring module 140, used for time sequence coding of the four sets of tire pressure value at the multiple predetermined time points to obtain an automobile tire pressure monitoring time sequence feature matrix; an automobile brake pad monitoring module 150, used for time sequence coding of the brake pad temperature value, the brake pad thickness value and the brake pad sound frequency value at the multiple predetermined time points to obtain an automobile brake pad monitoring time sequence feature matrix; an automobile monitoring feature fusion module 160, used for fusing the automobile electric drive monitoring time sequence feature matrix, the automobile battery monitoring time sequence feature matrix, the automobile tire pressure monitoring time sequence feature matrix and the automobile brake pad monitoring time sequence feature matrix to obtain an automobile monitoring feature matrix; an automobile fault monitoring result generation module 170, used for obtaining an automobile fault monitoring result based on the automobile monitoring feature matrix.

[0034] Specifically, in the technical solution of the present application, first, the driving motor speed value, the driving motor vibration frequency value, the battery pack voltage value, the battery pack current value, the battery pack internal resistance value, the battery pack temperature value, the four sets of tire pressure values, the brake pad temperature value, the brake pad thickness value, and the brake pad sound frequency value at multiple predetermined time points within a predetermined time period are obtained. It should be understood that the driving motor speed value reflects the running state of the motor, and abnormal speed may indicate mechanical failure, controller failure, or other problems of the motor; the driving motor vibration frequency value is related to problems such as motor bearing wear, imbalance, or mechanical component loosening, and is an important indicator of motor health; battery pack voltage fluctuations or abnormalities indicate inconsistent performance between battery cells, battery aging, or charging / discharging system problems; battery pack current abnormalities indicate internal short circuits, open circuits, or other problems; increased battery pack internal resistance is usually related to battery aging, active material reduction, or electrolyte failure; battery pack temperature abnormalities indicate battery overheating and thermal runaway, which are important safety monitoring parameters; insufficient or excessive tire pressure can affect tire grip, wear, and safety, and is an important part of automobile fault monitoring; high brake pad temperature indicates frequent use of the braking system, severe brake pad wear, or brake caliper failure; brake pad thickness reduction to a certain extent requires replacement, otherwise it will affect braking effectiveness and safety; abnormal brake pad sound indicates uneven brake pad wear, caliper loosening, or brake disc deformation. In general, these indicators are closely related to automobile faults, and by monitoring and analyzing the changes in these indicators in real time, potential fault risks can be detected in a timely manner to prevent the occurrence or expansion of faults, thereby ensuring the normal and safe operation of new energy vehicles.

[0035] In the embodiments of the present application, one implementation of obtaining the driving motor speed value, the driving motor vibration frequency value, the battery pack voltage value, the battery pack current value, the battery pack internal resistance value, the battery pack temperature value, the four sets of tire pressure values, the brake pad temperature value, the brake pad thickness value, and the brake pad sound frequency value at multiple predetermined time points within a predetermined time period can be: the driving motor speed value and the driving motor vibration frequency value can be obtained by collecting data through a speed sensor and a vibration sensor, respectively; the battery pack voltage value, the battery pack current value, the battery pack internal resistance value, and the battery pack temperature value can be obtained by collecting data through a voltage sensor, a current sensor, an internal resistance sensor, and a temperature sensor, respectively; tire pressure data can be obtained by collecting data through a wireless tire pressure sensor; the brake pad temperature value, the brake pad thickness value, and the brake pad sound frequency value can be obtained by collecting data through a temperature sensor, a thickness sensor, and a sound sensor, respectively; the vehicle control unit serves as the center of data collection, collects data from various sensors, and performs preliminary processing, and transmits the data to the cloud server through the Internet of Things communication module; the cloud server receives data from the vehicle and stores it in the database, and simultaneously analyzes and processes the collected operating data to monitor automobile faults.

[0036] Figure 2 This is a block diagram of the vehicle electric drive monitoring module in an IoT-based vehicle fault monitoring and diagnosis system according to an embodiment of this application. Figure 2 As shown, the vehicle electric drive monitoring module 120 includes: a vehicle electric drive monitoring related data structuring unit 121, used to obtain a vehicle electric drive monitoring time sequence input matrix by structuring the drive motor speed values ​​and drive motor vibration frequency values ​​at multiple predetermined time points according to the time dimension; and a vehicle electric drive monitoring time sequence feature extraction unit 122, used to obtain the vehicle electric drive monitoring time sequence feature matrix by passing the vehicle electric drive monitoring time sequence input matrix through a vehicle fault monitoring parameter time sequence feature encoder based on dilated convolution kernels.

[0037] Faults in automotive electric drive systems typically manifest as abnormal changes in motor speed and vibration frequency. To facilitate the analysis of collected automotive electric drive monitoring data and identify potential faults in the system, it is necessary to assemble the drive motor speed and vibration frequency values ​​at multiple predetermined time points into a time-series input matrix. By integrating the drive motor-related data into a single matrix, the efficiency of matrix operations can be leveraged for batch processing, thereby improving the efficiency of data analysis.

[0038] To extract useful features for fault identification from the original automotive electric drive monitoring time-series input matrix, such as periodic patterns, abnormal fluctuations, and trend changes in drive motor speed and vibration frequency—essential indicators for determining vehicle malfunctions—the automotive electric drive monitoring time-series input matrix needs to be processed by a dilated convolution kernel-based automotive fault monitoring parameter time-series feature encoder to obtain an automotive electric drive monitoring time-series feature matrix. In this application, the dilated convolution kernel-based automotive fault monitoring parameter time-series feature encoder is a convolutional neural network model using dilated convolution kernels. Dilated convolution, by introducing holes (i.e., zero values ​​in the convolution kernel), can increase the receptive field without increasing the number of parameters or computational cost. This means that dilated convolutional neural networks can capture data dependencies over a longer time span, thus more accurately extracting features from the automotive electric drive monitoring time-series input matrix. Furthermore, because dilated convolutional neural networks can capture input data dependencies over a longer time span, they exhibit better robustness to noise and outliers in the data. This means that in practical automotive fault monitoring and diagnosis applications, even if the input data contains some noise or abnormal fluctuations, the dilated convolutional neural network can still accurately extract useful features.

[0039] In this embodiment, one possible implementation of obtaining the vehicle electric drive monitoring time-series feature matrix by passing the vehicle fault monitoring parameter time-series feature encoder based on dilated convolution kernels through the vehicle fault monitoring parameter time-series feature encoder is as follows: During the forward propagation of each layer of the vehicle fault monitoring parameter time-series feature encoder based on dilated convolution kernels, the input data is processed as follows: The input data is convolved using the convolution units of the vehicle fault monitoring parameter time-series feature encoder based on dilated convolution kernels to obtain a convolutional feature map; the convolution units of each layer of the vehicle fault monitoring parameter time-series feature encoder based on dilated convolution kernels are used to perform pooling processing along the channel dimension on the convolutional feature map to obtain a pooled feature map; the activation units of each layer of the vehicle fault monitoring parameter time-series feature encoder based on dilated convolution kernels are used to perform nonlinear activation on the feature values ​​at each position in the pooled feature map to obtain an activation feature map; wherein, the output of the last layer of the vehicle fault monitoring parameter time-series feature encoder based on dilated convolution kernels is the vehicle electric drive monitoring time-series feature matrix.

[0040] Figure 3 This is a block diagram of a vehicle battery monitoring module in an IoT-based vehicle fault monitoring and diagnostic system according to an embodiment of this application. Figure 3 As shown, the vehicle battery monitoring module 130 includes: a vehicle battery monitoring related data structuring unit 131, used to arrange the battery pack voltage value, battery pack current value, battery pack internal resistance value, and battery pack temperature value at multiple predetermined time points according to the time dimension to obtain a vehicle battery monitoring time sequence input matrix; and a vehicle battery monitoring time sequence feature extraction unit 132, used to pass the vehicle battery monitoring time sequence input matrix through the vehicle fault monitoring parameter time sequence feature encoder based on the dilated convolution kernel to obtain the vehicle battery monitoring time sequence feature matrix.

[0041] Battery pack performance parameters (such as voltage, current, internal resistance, and temperature) change over time. These changes contain important information about the battery pack's operating status. To reveal the performance trends and potential faults of automotive battery packs, the battery pack voltage, current, internal resistance, and temperature values ​​at multiple predetermined time points can be arranged along a time dimension to obtain an automotive battery monitoring time series input matrix. This time series data, arranged along a time dimension, facilitates time series analysis. Each row in the automotive battery monitoring time series input matrix represents battery pack monitoring data at a single time point, and each column represents a specific battery pack monitoring indicator.

[0042] By analyzing the abnormal fluctuations or trend changes of relevant data in the automobile battery monitoring time series input matrix, the abnormality of the battery pack performance parameters can be found in time, and then the potential failure can be warned. For example, the sudden increase of the battery pack resistance means that there is a failure such as plate sulfuration, active material shedding, etc. in the battery pack. Therefore, in the technical solution of the present application, the automobile battery monitoring time series input matrix needs to be obtained through the automobile fault monitoring parameter time series feature encoder based on the hollow convolution kernel to obtain the automobile battery monitoring time series feature matrix. It can be understood that the hollow convolution kernel can capture the time-dependent features in the battery monitoring data by applying the convolution operation at different time steps, which can reflect the trend and periodicity of the battery performance parameters changing with time, and thus be beneficial to identifying the battery performance degradation and failure precursors.

[0043] Figure 4 The block diagram of the automobile tire monitoring module in the automobile fault monitoring and diagnosis system based on the Internet of Things according to the embodiment of the present application. As shown in Figure 4 The automobile tire monitoring module 140 includes: an automobile tire monitoring related data structuring unit 141, configured to arrange four groups of tire pressure values at a plurality of predetermined time points according to the time dimension and the tire dimension to obtain an automobile tire pressure monitoring time series input matrix; and an automobile tire monitoring time series feature extraction unit 142, configured to obtain the automobile tire pressure monitoring time series feature matrix by inputting the automobile tire pressure monitoring time series input matrix into the automobile fault monitoring parameter time series feature encoder based on the hollow convolution kernel.

[0044] In order to arrange the original tire pressure data collected by the wireless tire pressure sensor into a form convenient for subsequent analysis and processing, the four groups of tire pressure values at a plurality of predetermined time points can be arranged according to the time dimension and the tire dimension to obtain an automobile tire pressure monitoring time series input matrix. After arranging the tire pressure data into a time series input matrix, the time series analysis method can be conveniently applied to analyze the abnormal changes of the tire pressure, thereby realizing the diagnosis of the tire failure. Moreover, by considering the time dimension and the tire dimension at the same time, the change of the automobile tire pressure can be more comprehensively understood, and the accuracy and reliability of the automobile fault detection can be improved.

[0045] The original automobile tire pressure monitoring time sequence input matrix contains a large number of data points, which are detailed but often contain redundant information and noise. In order to effectively extract the key features of the tire pressure data and reduce the data dimension, the automobile tire pressure monitoring time sequence input matrix needs to be input into the automobile fault monitoring parameter time sequence feature encoder based on the hollow convolution kernel to obtain an automobile tire pressure monitoring time sequence feature matrix. It should be understood that the tire pressure data is time series data, and its trend and periodicity are crucial for judging the tire state. The hollow convolution kernel can expand the receptive field of the convolution kernel without increasing the computational complexity, so as to better capture the time sequence features in the tire pressure data. Compared with the traditional tire pressure monitoring method which only focuses on whether the absolute value of the tire pressure exceeds the preset range, the time sequence characteristics of the tire pressure data are ignored. Through the time sequence feature encoder based on the hollow convolution kernel, the tire pressure data can be more comprehensively analyzed, thereby improving the accuracy of automobile fault diagnosis.

[0046] Figure 5 A block diagram of an automobile brake pad monitoring module in the automobile fault monitoring and diagnosis system based on the Internet of Things according to an embodiment of the present application. As shown in Figure 5 The automobile brake pad monitoring module 150 includes an automobile brake pad monitoring related data structuring unit 151 for arranging the brake pad temperature values, brake pad thickness values, and brake pad sound frequency values at the plurality of predetermined time points according to the time dimension to obtain an automobile brake pad monitoring time sequence input matrix; and an automobile brake pad monitoring time sequence feature extraction unit 152 for inputting the automobile brake pad monitoring time sequence input matrix into the automobile fault monitoring parameter time sequence feature encoder based on the hollow convolution kernel to obtain the automobile brake pad monitoring time sequence feature matrix.

[0047] The brake pad is an important component in the automobile braking system, and its performance state changes with the use of the vehicle. Parameters such as temperature, thickness, and sound frequency can reflect the working state of the brake pad and its trend. In order to intuitively show the performance state of the brake pad at different time points and capture its dynamic change process, the brake pad temperature values, brake pad thickness values, and brake pad sound frequency values at the plurality of predetermined time points need to be arranged according to the time dimension to obtain an automobile brake pad monitoring time sequence input matrix. It should be understood that the brake pad may appear abnormal conditions such as wear and overheating during use, and these abnormal conditions often show specific patterns in the time dimension. For example, excessive wear of the brake pad will cause the thickness value to gradually decrease, and overheating will cause the temperature value to abnormally increase. By arranging the monitoring data according to the time dimension, these abnormal patterns can be more easily identified, providing a basis for subsequent fault warning and diagnosis. Each row of the automobile brake pad monitoring time sequence input matrix arranged is the automobile brake pad monitoring data at a time point, and each column is a specific automobile brake pad monitoring index.

[0048] The related data of the automobile brake pad at the current time point often has a dependency relationship with the data at multiple past time points. In order to capture the long-range dependency relationship in the automobile brake pad data, enable the model to understand the evolution law of the brake pad performance state over time, and further more accurately predict and diagnose faults, in the technical solution of the present application, the automobile brake pad monitoring time sequence input matrix is processed through the automobile fault monitoring parameter time sequence feature encoder based on the hollow convolution kernel to obtain an automobile brake pad monitoring time sequence feature matrix. The automobile brake pad monitoring time sequence feature matrix obtained through the automobile fault monitoring parameter time sequence feature encoder based on the hollow convolution kernel can represent the performance state of the brake pad and its change trend. When these feature information is abnormal, such as accelerated wear rate, continuously rising temperature, changed noise characteristics, etc., it can be judged that there is a fault hidden danger in the automobile brake system.

[0049] The power driving system, the battery system, the tire system and the brake system are crucial components in the automobile, and their states directly affect the overall performance and safety of the automobile. Considering that the running state of the automobile involves the cooperative work of multiple systems and components, in order to realize comprehensive monitoring and evaluation of multiple key systems of the automobile, so as to more accurately grasp the overall state of the automobile, in the technical solution of the present application, the automobile power driving monitoring time sequence feature matrix, the automobile battery monitoring time sequence feature matrix, the automobile tire pressure monitoring time sequence feature matrix and the automobile brake pad monitoring time sequence feature matrix are fused to obtain an automobile monitoring feature matrix. The fused automobile monitoring feature matrix contains information of multiple aspects such as power driving, battery, tire and brake system, which complement and confirm each other, providing a comprehensive basis for judging whether there is an obvious fault hidden danger in the automobile. Through fusion, it is conducive to improving the sensitivity of the fault diagnosis system to identify automobile fault hidden dangers. Even in the case that the data change of a single system is not obvious, but after combining with the data of other systems, potential fault risks may be revealed.

[0050] Figure 6 The block diagram of the automobile fault monitoring result generation module in the automobile fault monitoring and diagnosis system based on Internet of Things according to the embodiment of the present application. As shown in Figure 6As shown, the automobile fault monitoring result generation module 170 comprises: an automobile monitoring feature enhancement unit 171, configured to pass the automobile monitoring feature matrix through a spatial attention mechanism-based automobile monitoring feature enhancer to obtain an automobile monitoring feature enhanced matrix; an automobile monitoring feature matrix unfolding unit 172, configured to unfold the automobile monitoring feature enhanced matrix into an automobile monitoring feature enhanced vector; an automobile monitoring feature optimization unit 173, configured to perform backward propagation driven feature offset adjustment on the automobile monitoring feature enhanced vector to obtain an optimized automobile monitoring feature enhanced vector; and an automobile monitoring feature classification unit 174, configured to pass the optimized automobile monitoring feature enhanced vector through a classifier-based automobile fault monitoring result generator to obtain a classification result, which is used to indicate whether the automobile has a fault hidden danger at the current time point.

[0051] Although the automobile monitoring feature matrix obtained after fusion is comprehensive, its expression capability may be reduced due to information redundancy or interaction between features. In order to remove the redundant information in the feature matrix after fusion, strengthen the key features, and thus enhance the expression capability of the feature matrix, so that it better reflects the actual state of the automobile, the automobile monitoring feature matrix needs to be passed through a spatial attention mechanism-based automobile monitoring feature enhancer to obtain an automobile monitoring feature enhanced matrix. The spatial attention mechanism-based automobile monitoring feature enhancer in the present application is a convolutional neural network model using a spatial attention mechanism. It can be understood that the spatial attention mechanism can automatically learn and focus on the area features in the feature matrix that are more critical for fault monitoring. In the feature matrix after fusion, although the monitoring data of multiple systems are included, not all features are equally important for the analysis of the current fault. Through the spatial attention mechanism, the weights of different features can be dynamically adjusted, so that those features more related to the fault hidden danger occupy a larger proportion in the subsequent analysis. In this way, the possible fault features can be more prominent in the feature enhanced matrix, thereby accelerating the fault identification process.

[0052] In an embodiment of the present application, one implementation of the vehicle monitoring feature matrix through the vehicle monitoring feature enhancer based on the spatial attention mechanism to obtain a vehicle monitoring feature enhancement matrix is that: using each layer of the vehicle monitoring feature enhancer based on the spatial attention mechanism, in the forward transmission of the layer, respectively performs the following on the input data: performing convolution processing based on a convolution kernel on the input data to obtain a convolution feature map; passing the convolution feature map through a spatial attention unit to obtain a spatial attention map; calculating the point-by-point multiplication of the convolution feature map and the spatial attention map by position to obtain a spatial attention feature map; inputting the spatial attention feature map into a nonlinear activation unit to obtain an activated feature map; wherein the input of the first layer of the vehicle monitoring feature enhancer is the vehicle monitoring feature matrix, and the output of the last layer of the vehicle monitoring feature enhancer is the vehicle monitoring feature enhancement matrix.

[0053] Considering that the vehicle monitoring feature enhancement matrix has a high dimension, which is not conducive to subsequent data processing, in the technical solution of the present application, the vehicle monitoring feature enhancement matrix needs to be unfolded into a vehicle monitoring feature enhancement vector. Through the unfolding operation, the dimensionality reduction of feature data can be realized, and the reduced data can reduce the computational complexity and storage requirements, thereby being able to speed up the running speed of the vehicle fault diagnosis algorithm.

[0054] In particular, considering that the present application relies on various sensors to collect monitoring data of the vehicle, these sensors will produce errors due to manufacturing defects, aging or environmental factors (such as temperature, humidity), and in the data acquisition process, they will also be affected by factors such as electromagnetic interference, signal attenuation, etc. These factors will introduce noise into the collected data, thereby causing local feature disturbances in the vehicle monitoring feature enhancement vector formed by the final encoding. Since in actual application, the vehicle fault monitoring system needs to be operated stably for a long time, in order to ensure that reliable vehicle fault monitoring can be provided under various conditions, in the technical solution of the present application, the vehicle monitoring feature enhancement vector needs to be adjusted for feature offset driven by back propagation to obtain an optimized vehicle monitoring feature enhancement vector.

[0055] Specifically, in the embodiment of the present application, the backward propagation driven feature shift adjustment on the vehicle monitoring feature enhancement vector is performed to obtain an optimized vehicle monitoring feature enhancement vector, including: determining a first weight matrix and a second weight matrix before and after each iteration update of the classifier; performing matrix multiplication on the first weight matrix and the second weight matrix and the vehicle monitoring feature enhancement vector respectively to obtain a first vehicle monitoring modulation feature vector and a second vehicle monitoring modulation feature vector; calculating the positional difference between the first vehicle monitoring modulation feature vector and the second vehicle monitoring modulation feature vector to obtain a vehicle monitoring shift information representation vector; calculating the F-norm of the positional mean vector between the first vehicle monitoring modulation feature vector and the second vehicle monitoring modulation feature vector as a shift compensation scaling factor; performing linear scaling on the vehicle monitoring shift information representation vector by the shift compensation scaling factor to obtain a scaled vehicle monitoring shift information representation vector, and inputting the scaled vehicle monitoring shift information representation vector into a Sigmoi d activation function to obtain a vehicle monitoring backward propagation shift compensation representation vector; calculating the positional point multiplication between the vehicle monitoring backward propagation shift compensation representation vector and the vehicle monitoring feature enhancement vector to obtain the optimized vehicle monitoring feature enhancement vector.

[0056] In the embodiment of the present application, another expression manner of the backward propagation driven feature shift adjustment on the vehicle monitoring feature enhancement vector to obtain an optimized vehicle monitoring feature enhancement vector can be: processing the vehicle monitoring feature enhancement vector by the following formula to obtain the optimized vehicle monitoring feature enhancement vector; wherein the formula is:

[0057]

[0058] wherein M1 is the first weight matrix, M2 is the second weight matrix, V is the vehicle monitoring feature enhancement vector, represents matrix multiplication, represents positional point addition, and ||·||F represents the F-norm of a feature vector. F represents the F-norm of a feature vector, S represents the shift compensation scaling factor, and represents positional point multiplication. represents positional point subtraction, and sigmoid represents an activation function, and V' represents the optimized vehicle monitoring feature enhancement vector.

[0059] In order to enhance the robustness of the classifier to local perturbations in the input feature vector, the automobile monitoring feature enhancement vector is subjected to shift compensation based on the back propagation representation. The key to subjecting the automobile monitoring feature enhancement vector to shift compensation based on the back propagation representation lies in capturing the information of the change of the weight matrix during the training process, and using this information to generate an automobile monitoring shift information representation vector, and then generating an automobile monitoring back propagation shift compensation representation vector through scaling and nonlinear transformation (Sigmoid function). In this way, the automobile monitoring feature enhancement vector is subjected to local shift description by the distribution difference of different classification scenarios of the classifier during the training process, so as to not only quantify the displacement of the feature vector caused by the change of the weight matrix, but also generate a compensation vector according to the displacement information, which adjusts each element in the feature vector according to the influence degree of the perturbation, thereby reducing the influence of the perturbation on the classification result. In this way, the robustness of the classifier to local perturbations in the input data can be significantly improved, the stability and generalization ability of the classifier are enhanced, and by reducing the influence of local perturbations on the classification result, the classifier can still maintain high classification accuracy when facing noise or variant data.

[0060] Finally, the optimized automobile monitoring feature enhancement vector is subjected to a classifier-based automobile fault monitoring result generator to obtain a classification result, which is used to represent whether the automobile at the current time point has a fault hidden danger. It should be understood that the optimized automobile monitoring feature enhancement vector is high-dimensional and not easy to understand directly. In order to automatically convert complex monitoring feature data into easily understood fault classification results, the optimized automobile monitoring feature enhancement vector can be subjected to a classifier-based automobile fault monitoring result generator. As a machine learning model, the classifier can analyze and judge the input data and map it to different categories, that is, according to the information of the optimized automobile monitoring feature enhancement vector, whether the automobile at the current time point has a fault hidden danger is judged. When the classifier outputs a classification result indicating that there is a fault hidden danger, the system can remind the driver by issuing an alarm, so as to take timely measures to avoid the expansion of the fault or cause an accident. In this way, by analyzing real-time automobile key component data, potential faults can be detected earlier and the accuracy of fault diagnosis can be improved.

[0061] In summary, the automobile fault monitoring and diagnosis system 100 based on the Internet of Things based on the embodiments of the present application is illustrated, which first collects real-time running data of various components of the automobile through various sensors, then uses artificial intelligence-based data analysis technology to analyze the collected automobile component running data in real time, and then judges whether the automobile has a fault hidden danger. In this way, more accurate and efficient automobile fault monitoring can be achieved, thereby providing a strong guarantee for the safe operation of new energy vehicles.

[0062] As described above, the IoT-based vehicle fault monitoring and diagnosis system 100 according to embodiments of this application can be implemented in various terminal devices, such as servers for IoT-based vehicle fault monitoring and diagnosis. In one example, the IoT-based vehicle fault monitoring and diagnosis system 100 according to embodiments of this application can be integrated into a terminal device as a software module and / or hardware module. For example, the IoT-based vehicle fault monitoring and diagnosis system 100 can be a software module in the operating system of the terminal device, or it can be an application developed for the terminal device; of course, the IoT-based vehicle fault monitoring and diagnosis system 100 can also be one of many hardware modules of the terminal device.

[0063] Alternatively, in another example, the IoT-based vehicle fault monitoring and diagnostic system 100 and the terminal device can also be separate devices, and the IoT-based vehicle fault monitoring and diagnostic system 100 can be connected to the terminal device via wired and / or wireless networks, and transmit interactive information in accordance with an agreed data format.

[0064] Figure 7 This is a flowchart of an IoT-based vehicle fault monitoring and diagnosis method according to an embodiment of this application. Figure 7 As shown, the IoT-based vehicle fault monitoring and diagnosis method includes: S110, acquiring the drive motor speed value, drive motor vibration frequency value, battery pack voltage value, battery pack current value, battery pack internal resistance value, battery pack temperature value, four sets of tire pressure values, brake pad temperature value, brake pad thickness value, and brake pad sound frequency value at multiple predetermined time points within a predetermined time period; S120, performing time-series encoding on the drive motor speed value and drive motor vibration frequency value at the multiple predetermined time points to obtain a vehicle electric drive monitoring time-series feature matrix; S130, performing time-series encoding on the battery pack voltage value, battery pack current value, battery pack internal resistance value, and battery pack temperature value at the multiple predetermined time points to obtain a vehicle electric drive monitoring time-series feature matrix. S140: Obtain the vehicle battery monitoring time-series feature matrix; S150: Perform time-series encoding on the four sets of tire pressure values ​​at the multiple predetermined time points to obtain the vehicle tire pressure monitoring time-series feature matrix; S160: Perform time-series encoding on the brake pad temperature value, brake pad thickness value, and brake pad sound frequency value at the multiple predetermined time points to obtain the vehicle brake pad monitoring time-series feature matrix; S170: Fuse the vehicle electric drive monitoring time-series feature matrix, the vehicle battery monitoring time-series feature matrix, the vehicle tire pressure monitoring time-series feature matrix, and the vehicle brake pad monitoring time-series feature matrix to obtain the vehicle monitoring feature matrix; S180: Obtain the vehicle fault monitoring result based on the vehicle monitoring feature matrix.

[0065] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned Internet of Things-based automobile fault monitoring and diagnosis method have been described in detail above with reference to the description of the Internet of Things-based automobile fault monitoring and diagnosis system, and therefore the repeated description thereof will be omitted. Figures 1 to 6

[0066] In summary, the Internet of Things-based automobile fault monitoring and diagnosis method based on the embodiments of the present application is illustrated, which first collects real-time operation data of each component of the automobile through various sensors, then uses artificial intelligence-based data analysis technology to analyze the collected operation data of each component of the automobile in real time, and further judges whether the automobile has hidden faults. In this way, more accurate and efficient automobile fault monitoring can be achieved, thereby providing strong protection for the safe operation of new energy vehicles.​

Claims

1. An Internet of Things based automobile fault monitoring and diagnostic system characterized in that, Comprise: The automobile fault monitoring related data acquisition module is used for acquiring driving motor rotating speed values, driving motor vibration frequency values, battery pack voltage values, battery pack current values, battery pack internal resistance values, battery pack temperature values, four groups of tire pressure values, brake pad temperature values, brake pad thickness values and brake pad sound frequency values at a plurality of predetermined time points in a predetermined time period; The automobile power drive monitoring module is used for time series coding of the driving motor rotating speed values and the driving motor vibration frequency values at the plurality of predetermined time points to obtain an automobile power drive monitoring time series feature matrix; The automobile battery monitoring module is used for time series coding of the battery pack voltage values, the battery pack current values, the battery pack internal resistance values and the battery pack temperature values at the plurality of predetermined time points to obtain an automobile battery monitoring time series feature matrix; The automobile tire monitoring module is used for time series coding of the four groups of tire pressure values at the plurality of predetermined time points to obtain an automobile tire pressure monitoring time series feature matrix; The automobile brake pad monitoring module is used for time series coding of the brake pad temperature values, the brake pad thickness values and the brake pad sound frequency values at the plurality of predetermined time points to obtain an automobile brake pad monitoring time series feature matrix; The automobile monitoring feature fusion module is used for fusing the automobile power drive monitoring time series feature matrix, the automobile battery monitoring time series feature matrix, the automobile tire pressure monitoring time series feature matrix and the automobile brake pad monitoring time series feature matrix to obtain an automobile monitoring feature matrix; The automobile fault monitoring result generation module is used for obtaining an automobile fault monitoring result based on the automobile monitoring feature matrix; The automobile fault monitoring result generation module comprises: The automobile monitoring feature enhancement unit is used for obtaining an automobile monitoring feature enhancement matrix through an automobile monitoring feature enhancer based on a spatial attention mechanism; The automobile monitoring feature matrix unfolding unit is used for unfolding the automobile monitoring feature enhancement matrix into an automobile monitoring feature enhancement vector; The automobile monitoring feature optimization unit is used for performing feature bias adjustment driven by back propagation on the automobile monitoring feature enhancement vector to obtain an optimized automobile monitoring feature enhancement vector; The automobile monitoring feature classification unit is used for obtaining a classification result through an automobile fault monitoring result generator based on a classifier, the classification result being used for indicating whether the automobile at the current time point has a fault hidden danger; The automobile monitoring feature optimization unit is used for: Determining a first weight matrix and a second weight matrix of the classifier before and after each iteration update; Performing matrix multiplication of the first weight matrix and the second weight matrix with the automobile monitoring feature enhancement vector respectively to obtain a first automobile monitoring modulation feature vector and a second automobile monitoring modulation feature vector; Calculating a position difference between the first automobile monitoring modulation feature vector and the second automobile monitoring modulation feature vector to obtain an automobile monitoring shift information representation vector; Calculating an F-norm of a position mean vector between the first automobile monitoring modulation feature vector and the second automobile monitoring modulation feature vector as a shift compensation scaling factor; linearly scaling the vehicle monitoring shift information representation vector by the shift compensation scaling factor to obtain a scaled vehicle monitoring shift information representation vector, and inputting the scaled vehicle monitoring shift information representation vector into a Sigmoid activation function to obtain a vehicle monitoring back propagation shift compensation representation vector; calculating a point-wise multiplication of the vehicle monitoring back propagation shift compensation representation vector and the vehicle monitoring feature enhancement vector to obtain the optimized vehicle monitoring feature enhancement vector; wherein, the shift compensation based on back propagation representation is performed on the vehicle monitoring feature enhancement vector to capture the information of the change of the weight matrix during the training process, and these information is used to generate the vehicle monitoring shift information representation vector, and then the vehicle monitoring back propagation shift compensation representation vector is generated through scaling and nonlinear transformation, so that the feature local shift description is performed on the vehicle monitoring feature enhancement vector by supporting the distribution difference of different classification scenarios of the classifier during the training process, so as to not only quantify the displacement of the feature vector caused by the change of the weight matrix, but also generate a compensation vector according to the displacement information, and the compensation vector is used to adjust each element in the feature vector according to the influence degree of the disturbance, so as to reduce the influence of the disturbance on the classification result.

2. The Internet of Things based automobile failure monitoring diagnostic system as claimed in claim 1 wherein, The vehicle power drive monitoring module comprises: A vehicle power drive monitoring related data structuring unit is configured to arrange the drive motor speed values and drive motor vibration frequency values at the plurality of predetermined time points according to the time dimension to obtain a vehicle power drive monitoring time sequence input matrix. A vehicle power drive monitoring time sequence feature extraction unit is configured to pass the vehicle power drive monitoring time sequence input matrix through the automobile fault monitoring parameter time sequence feature encoder based on the hollow convolution kernel to obtain the vehicle power drive monitoring time sequence feature matrix.

3. The Internet of Things based automobile failure monitoring diagnostic system as claimed in claim 2 wherein, The vehicle battery monitoring module comprises: A vehicle battery monitoring related data structuring unit is configured to arrange the battery pack voltage values, battery pack current values, battery pack internal resistance values, and battery pack temperature values at the plurality of predetermined time points according to the time dimension to obtain a vehicle battery monitoring time sequence input matrix. A vehicle battery monitoring time sequence feature extraction unit is configured to pass the vehicle battery monitoring time sequence input matrix through the automobile fault monitoring parameter time sequence feature encoder based on the hollow convolution kernel to obtain the vehicle battery monitoring time sequence feature matrix.

4. The Internet of Things based automobile failure monitoring diagnostic system as claimed in claim 3, wherein, The vehicle tire monitoring module comprises: A vehicle tire monitoring related data structuring unit is configured to arrange the four groups of tire pressure values at the plurality of predetermined time points according to the time dimension and the tire dimension to obtain a vehicle tire pressure monitoring time sequence input matrix. A vehicle tire monitoring time sequence feature extraction unit is configured to pass the vehicle tire pressure monitoring time sequence input matrix through the automobile fault monitoring parameter time sequence feature encoder based on the hollow convolution kernel to obtain the vehicle tire pressure monitoring time sequence feature matrix.

5. The Internet of Things based automobile failure monitoring diagnostic system as claimed in claim 4, wherein, The vehicle brake pad monitoring module comprises: The automobile brake pad monitoring related data structuring unit is configured to arrange brake pad temperature values, brake pad thickness values, and brake pad sound frequency values at a plurality of predetermined time points according to a time dimension to obtain an automobile brake pad monitoring time series input matrix. The automobile brake pad monitoring time series feature extraction unit is configured to pass the automobile brake pad monitoring time series input matrix through the automobile fault monitoring parameter time series feature encoder based on a cavity convolution kernel to obtain an automobile brake pad monitoring time series feature matrix.

6. The Internet of Things based automobile failure monitoring diagnostic system as claimed in claim 5 wherein, The automobile fault monitoring parameter time series feature encoder based on a cavity convolution kernel is a convolutional neural network model using a cavity convolution kernel, and the automobile monitoring feature enhancer based on a spatial attention mechanism is a convolutional neural network model using a spatial attention mechanism.

7. A method for monitoring and diagnosing automobile faults based on Internet of Things, using the system for monitoring and diagnosing automobile faults based on Internet of Things as claimed in claim 1, wherein, The method comprises: obtaining driving motor speed values, driving motor vibration frequency values, battery pack voltage values, battery pack current values, battery pack internal resistance values, battery pack temperature values, four sets of tire pressure values, brake pad temperature values, brake pad thickness values, and brake pad sound frequency values at a plurality of predetermined time points within a predetermined time period; time series encoding the driving motor speed values and the driving motor vibration frequency values at the plurality of predetermined time points to obtain an automobile electric power driving monitoring time series feature matrix; time series encoding the battery pack voltage values, the battery pack current values, the battery pack internal resistance values, and the battery pack temperature values at the plurality of predetermined time points to obtain an automobile battery monitoring time series feature matrix; time series encoding the four sets of tire pressure values at the plurality of predetermined time points to obtain an automobile tire pressure monitoring time series feature matrix; time series encoding the brake pad temperature values, the brake pad thickness values, and the brake pad sound frequency values at the plurality of predetermined time points to obtain an automobile brake pad monitoring time series feature matrix; fusing the automobile electric power driving monitoring time series feature matrix, the automobile battery monitoring time series feature matrix, the automobile tire pressure monitoring time series feature matrix, and the automobile brake pad monitoring time series feature matrix to obtain an automobile monitoring feature matrix; obtaining an automobile fault monitoring result based on the automobile monitoring feature matrix.

Images