Vehicle fault diagnosis method and device based on multi-source data fusion and storage medium

Abstract

The application provides a vehicle fault diagnosis method and device based on multi-source data fusion and a storage medium. The method comprises: collecting real-time state data and real-time alarm data of a target part, obtaining external environment data and map navigation data; after data preprocessing, the multi-source data set target features are obtained by fusion, the target features extracted from the multi-source data set are input into the adaptive neural fuzzy inference system, the target features are mapped to multiple fuzzy sets, and the member function is used for fuzzification; the rule weight of each target feature is calculated to reflect the fuzzy strength; the neural network learning algorithm is used to train the inference system, the member function and the rule weight are adjusted, the fuzzy output is converted into a specific numerical value or a fault level in the defuzzification process by using the output member function, so that the health state of the part is judged. The application provides a fault diagnosis output with high interpretability, improves the accuracy, and can perform fault diagnosis in a wide range of vehicle operating environments.

Description

Technical Field

[0001] This application relates to the field of new energy vehicle technology, and in particular to a vehicle fault diagnosis method, device and storage medium based on multi-source data fusion. Background Technology

[0002] With the continuous development of modern transportation technology, vehicle systems are becoming increasingly complex. To ensure vehicle safety and reliability, fault diagnosis has become an indispensable part. Traditional fault diagnosis methods are usually based on a fixed model or expert experience, but with the increase in data volume and the growing complexity of vehicle systems, these methods may fail to accurately detect and diagnose faults.

[0003] Adaptive Neural Fuzzy Inference Systems (ANFIS), as a model integrating neural networks and fuzzy inference techniques, have demonstrated their effectiveness in multiple fields. However, their application in vehicle fault diagnosis remains an area to be explored, especially in conjunction with multi-source data fusion methods for fault diagnosis.

[0004] Existing technologies have not effectively collected and integrated multi-source vehicle data for vehicle fault diagnosis. Furthermore, traditional methods based on fixed thresholds or model-based judgments result in poor interpretability and low accuracy of vehicle fault diagnosis results, and they cannot accurately diagnose different types of faults under actual vehicle operating conditions. Summary of the Invention

[0005] In view of this, embodiments of this application provide a vehicle fault diagnosis method, device and storage medium based on multi-source data fusion to solve the problems of poor interpretability of vehicle fault diagnosis results, low accuracy of fault diagnosis results and inability to match different types of fault diagnosis under actual vehicle operating conditions in the prior art.

[0006] A first aspect of this application provides a vehicle fault diagnosis method based on multi-source data fusion, comprising: collecting real-time status data and real-time alarm data of target components; collecting external environment data using sensors installed on the vehicle; and acquiring map navigation data of the vehicle; performing preprocessing operations on the real-time status data, real-time alarm data, external environment data, and map navigation data of the target components; fusing the preprocessed data to obtain a fused multi-source dataset; extracting features from the multi-source dataset to obtain target features for vehicle fault detection; and inputting the target features into a predetermined adaptive neurofuzzy inference system, wherein the adaptive neurofuzzy inference system includes expert-based... The system identifies multiple fuzzy rules; utilizes an adaptive neural fuzzy inference system to map target features to different fuzzy sets, and uses input member functions to fuzzify the target features; for each fuzzy rule in the fuzzy set, it calculates the rule weight corresponding to each target feature, which characterizes the fuzziness intensity corresponding to the target feature; it trains the adaptive neural fuzzy inference system using a neural network learning algorithm and a training dataset to adjust the member functions and rule weights in the system; during the defuzzification process, it uses output member functions to convert the fuzzy output into specific numerical values ​​or fault levels, and judges the health status of the target component based on the output of the adaptive neural fuzzy inference system.

[0007] A second aspect of this application provides a vehicle fault diagnosis device based on multi-source data fusion, comprising: a data acquisition module configured to acquire real-time status data and real-time alarm data of target components, acquire external environmental data using sensors installed on the vehicle, and obtain vehicle map navigation data; a preprocessing module configured to perform preprocessing operations on the real-time status data, real-time alarm data, external environmental data, and map navigation data of the target components, fuse the preprocessed data, and obtain a fused multi-source dataset; and a feature extraction module configured to extract features from the multi-source dataset to obtain target features for vehicle fault detection, and input the target features into a predetermined adaptive neurofuzzy inference system, wherein the adaptive neurofuzzy inference system includes features defined based on expert knowledge. The system includes: a fuzzy rules module; a fuzzy processing module configured to map target features to different fuzzy sets using an adaptive neural fuzzy inference system and to fuzzify the target features using input member functions; a calculation module configured to calculate the rule weight corresponding to each target feature for each fuzzy rule in the fuzzy set, where the rule weight represents the fuzziness intensity corresponding to the target feature; a training module configured to train the adaptive neural fuzzy inference system using a neural network learning algorithm and a training dataset to adjust the member functions and rule weights in the adaptive neural fuzzy inference system; and an output module configured to convert the fuzzy output into specific numerical values ​​or fault levels using output member functions during the defuzzification process, and to determine the health status of the target component based on the output of the adaptive neural fuzzy inference system.

[0008] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the above-described method.

[0009] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the above-described method.

[0010] The above-described technical solutions adopted in the embodiments of this application can achieve the following beneficial effects:

[0011] By collecting real-time status data and real-time alarm data of target components, using sensors installed on the vehicle to collect external environmental data, and acquiring vehicle map navigation data, the system performs preprocessing operations on the real-time status data, real-time alarm data, external environmental data, and map navigation data of the target components. The preprocessed data is then fused to obtain a fused multi-source dataset. Feature extraction is performed on the multi-source dataset to obtain target features for vehicle fault detection. These target features are input into a predetermined adaptive neurofuzzy inference system, which includes multiple fuzzy rules defined based on expert knowledge. The adaptive neurofuzzy inference system maps the target features to different fuzzy sets, and uses input member functions to fuzzify the target features. For each fuzzy rule in the fuzzy set, the rule weight corresponding to each target feature is calculated; the rule weight is used to characterize the fuzziness intensity corresponding to the target feature. The adaptive neurofuzzy inference system is trained using a neural network learning algorithm and a training dataset to adjust the member functions and rule weights within the system. During the defuzzification process, the output member function is used to convert the fuzzy output into a specific numerical value or fault level. Based on the output of the adaptive neurofuzzy inference system, the health status of the target components is determined. The vehicle fault diagnosis results output in this application are highly interpretable, improving the accuracy of the fault diagnosis results and covering fault diagnosis for more vehicles under operation. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a flowchart illustrating the vehicle fault diagnosis method based on multi-source data fusion provided in the embodiments of this application;

[0014] Figure 2 This is a schematic diagram of the structure of the vehicle fault diagnosis device based on multi-source data fusion provided in the embodiments of this application;

[0015] Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0016] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0017] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0018] The vehicle fault diagnosis method and apparatus based on multi-source data fusion provided in this application will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] Figure 1 This is a flowchart illustrating the vehicle fault diagnosis method based on multi-source data fusion provided in the embodiments of this application. Figure 1 The vehicle fault diagnosis method based on multi-source data fusion can be executed by the server.

[0020] like Figure 1 As shown, this vehicle fault diagnosis method based on multi-source data fusion may specifically include:

[0021] S101 collects real-time status data and real-time alarm data of target components, collects external environmental data using sensors installed on the vehicle, and obtains the vehicle's map navigation data.

[0022] S102, perform preprocessing operations on the real-time status data, real-time alarm data, external environment data and map navigation data of the target components, and fuse the preprocessed data to obtain the fused multi-source dataset;

[0023] S103, extract features from the multi-source dataset to obtain target features for vehicle fault detection, and input the target features into a predetermined adaptive neurofuzzy inference system, wherein the adaptive neurofuzzy inference system includes a variety of fuzzy rules defined based on expert knowledge;

[0024] S104 utilizes an adaptive neural fuzzy inference system to map target features to different fuzzy sets and uses input member functions to fuzzify the target features;

[0025] S105, For each fuzzy rule in the fuzzy set, calculate the rule weight corresponding to each target feature. The rule weight is used to characterize the fuzzy intensity corresponding to the target feature.

[0026] S106, The adaptive neurofuzzy reasoning system is trained using a neural network learning algorithm and a training dataset in order to adjust the member functions and rule weights in the adaptive neurofuzzy reasoning system;

[0027] S107, during the defuzzification process, the output member function is used to convert the fuzzy output into a specific value or fault level, and the health status of the target component is determined based on the output of the adaptive neural fuzzy inference system.

[0028] An Adaptive Neuro-Fuzzy Inference System (ANFIS) is an intelligent system that combines fuzzy logic systems and neural networks. ANFIS leverages the interpretive power of fuzzy logic systems and the learning capabilities of neural networks to achieve adaptive learning and adjustment of fuzzy systems.

[0029] The working process of an adaptive neural fuzzy inference system (ANFIS) includes the following steps: First, ANFIS receives raw input data and transforms it into fuzzy values ​​or membership degrees using input membership functions. This process is called fuzzification. The fuzzified data is then fed into a predefined set of fuzzy rules. These rules are evaluated based on the membership degrees of the input data, generating an output value for each rule. Each fuzzy rule's output has an associated weight. These weights are typically determined based on the membership degrees of the input data and the structure of the rules. The weighted rule outputs are then aggregated or summed to produce a total fuzzy output. Finally, ANFIS uses output membership functions to transform the fuzzy output into a specific numerical value; this process is called defuzzification. One of the characteristics of ANFIS is its ability to learn and adjust. Based on training data, the neural network algorithm (such as backpropagation) adjusts the system, optimizing the parameters of the membership functions and the rule weights to reduce prediction errors.

[0030] In some embodiments, real-time status data and real-time alarm data of the target components are collected, external environmental data is collected using sensors installed on the vehicle, and map navigation data of the vehicle is obtained, including:

[0031] The system utilizes a vehicle-mounted status monitoring system to monitor the real-time status data generated by target components. When an abnormal state or parameter exceeding a predetermined range is detected, real-time alarm data is generated. The system also uses environmental sensors on the vehicle to collect real-time data on the external environment in which the vehicle is located, obtains map navigation data from the vehicle's navigation system, and retrieves weather data for the expected driving route from a pre-defined application.

[0032] Specifically, the new energy vehicle (hereinafter referred to as the vehicle) of this application is equipped with an advanced condition monitoring system, which is integrated into various key components of the vehicle, such as the engine, braking system, tires, and battery. These monitoring points can monitor the working status of these components in real time, such as engine speed, braking system pressure, tire pressure, and battery charge. When these data show abnormalities or any parameter exceeds the preset safety range, the monitoring system will immediately generate and send real-time alarm data.

[0033] Meanwhile, the vehicle is also equipped with multiple environmental sensors, distributed at the front, rear, left, right, and top of the vehicle. These sensors can monitor the vehicle's external environmental conditions in real time, such as temperature, humidity, rain detection, and road conditions. This data is crucial for determining the vehicle's operating status in specific environments.

[0034] In addition, the vehicle is equipped with a high-precision navigation system. This system not only provides current location information but also real-time map navigation data such as upcoming road conditions, traffic flow, and traffic accidents. This data helps the vehicle prepare in advance and avoid potential risks.

[0035] To better assess the vehicle's operational status on upcoming road sections, this application also integrates a weather forecast application. This application provides real-time and forecast weather data for the vehicle's expected route, such as temperature, precipitation, wind speed, and wind direction.

[0036] In some embodiments, preprocessing operations are performed on the real-time status data, real-time alarm data, external environment data, and map navigation data of the target component, including:

[0037] The raw data includes real-time status data of the target components, real-time alarm data, external environment data, map navigation data, and weather data.

[0038] Data cleaning is performed on the raw data to detect and remove missing, duplicate, or erroneous values; scaling is performed on different types of raw data to unify the raw data from different dimensions; and anomaly detection is performed on the raw data using an autoencoder model to identify anomalous data.

[0039] Specifically, in vehicle fault diagnosis systems, to ensure data quality and validity, all collected data, including real-time status data of target components, real-time alarm data, external environmental data, map navigation data, and weather data, are first identified as raw data. This raw data is obtained directly from various sensors and systems and has not undergone any processing.

[0040] Furthermore, the raw data must first undergo data cleaning. Data cleaning primarily ensures the integrity, consistency, and accuracy of the data. During the data cleaning process, the system automatically detects and deletes missing, duplicate, or erroneous values ​​in the raw data. For example, if a sensor sends two identical data points, this duplicate data will be deleted; similarly, if a sensor sends data that is significantly outside the normal range, such as a temperature sensor suddenly reporting a temperature of -200℃, this data will also be marked as erroneous and deleted.

[0041] Furthermore, to ensure data consistency in subsequent processing, the system scales different types of raw data. This is because data from different sensors or systems may have different scales or units. For example, engine speed may be measured in "revolutions per minute," while temperature may be measured in "°C." To unify this data, the system uses methods such as max-min scaling and Z-score normalization to transform the data to the same scale, such as the range [0,1].

[0042] In one example, to further ensure data reliability, an autoencoder model is used to detect anomalies in the raw data. An autoencoder is a neural network model that learns the features of data and reconstructs it. In this application, the autoencoder is trained to recognize normal data patterns. When the input data deviates from this pattern, the autoencoder's reconstruction error increases, and the system treats this data as anomalous and marks it.

[0043] In some embodiments, the preprocessed data is fused to obtain a fused multi-source dataset, including:

[0044] Based on the fusion weights corresponding to the original data in different dimensions, the original data is weighted to obtain a weighted multi-source dataset.

[0045] Alternatively, feature fusion can be performed on raw data from different dimensions, and multiple raw data can be fused together to obtain a fused multi-source dataset.

[0046] Alternatively, the original data can be classified, and the categories of the classified original data can be fused based on the classification results to obtain a fused multi-source dataset.

[0047] Specifically, considering that the importance of raw data from different dimensions for fault diagnosis may vary, different weights can be assigned to each data dimension. For example, for engine fault diagnosis, engine temperature and oil pressure data may be more important than external temperature or map navigation data. Therefore, higher weights can be assigned to engine temperature and oil pressure data. By combining these weights through a weighted average, the data dimensions are weighted and fused to obtain a weighted multi-source dataset.

[0048] In another example, besides direct weighted fusion, feature fusion can be performed on the raw data from different dimensions. Feature fusion involves combining or transforming different data features into new features, thus reflecting the characteristics of the data from different perspectives. For example, engine temperature and RPM can be fused into a single feature called "engine efficiency." Through feature fusion, a new multi-source dataset that integrates multiple raw data sources can be obtained.

[0049] In another example, considering that the original data may contain multiple categories, such as sensor source, data type (e.g., temperature, pressure, speed), or other factors, the data can first be categorized according to these categories. Then, based on the characteristics and importance of each category, an appropriate fusion method (e.g., weighted fusion, feature fusion, etc.) is selected for each category. Finally, these fused category data are further integrated to obtain the final fused multi-source dataset.

[0050] In some embodiments, an adaptive neural fuzzy inference system is used to map target features to different fuzzy sets, and input membership functions are used to fuzzify the target features, including:

[0051] Based on the target features, define the corresponding fuzzy variables, and define a fuzzy set for each fuzzy variable. Select a membership function to describe each fuzzy set so that each target feature obtains a membership value on the corresponding fuzzy set.

[0052] For each target feature, its membership value on each fuzzy set is calculated using the selected input member function in order to convert each target feature into a fuzzy value. The calculated fuzzy value is stored in a fuzzy matrix for fuzzy inference and defuzzification operations.

[0053] Specifically, for each target feature in vehicle fault diagnosis, such as engine temperature, engine speed, or oil pressure, corresponding fuzzy variables are first defined. For example, engine temperature can be defined as three fuzzy variables: "low," "medium," and "high." Subsequently, a fuzzy set is defined for each fuzzy variable; for example, "low" temperature can correspond to a temperature range (e.g., 0-50℃), and "medium" temperature can correspond to another range (e.g., 50-80℃), and so on.

[0054] Furthermore, based on the characteristics of each fuzzy set, an appropriate membership function is selected to describe it. Commonly used membership functions include triangular, trapezoidal, or Gaussian functions. For example, for a "low" temperature fuzzy set, a triangular membership function that increases in the range of 0-50℃ might be chosen; while for a "medium" temperature fuzzy set, a Gaussian function centered at 65℃ might be chosen.

[0055] Furthermore, when real-time target feature data is generated during vehicle operation, such as an engine temperature of 55°C, the membership values ​​of the target feature on each fuzzy set can be calculated using the previously selected input membership function. For example, under the triangular membership function of the "low" temperature fuzzy set, the membership value of 55°C may be 0.5; while under the Gaussian function of the "medium" temperature fuzzy set, its membership value may be 0.7.

[0056] Furthermore, the calculated fuzzy values ​​of each target feature are stored in a fuzzy matrix. For example, the rows of the fuzzy matrix represent different target features, such as engine temperature and RPM, while the columns represent various fuzzy variables, such as "low," "medium," and "high." The corresponding matrix elements are the membership values ​​of the target feature under the corresponding fuzzy variables.

[0057] Using the method described in the above embodiments, each target feature is converted into a fuzzy value and integrated into a fuzzy matrix. This fuzzy matrix provides the basis for subsequent fuzzy inference and defuzzification operations, thereby helping to determine the health status and potential faults of various vehicle components.

[0058] In some embodiments, for each fuzzy rule in the fuzzy set, the rule weight corresponding to each target feature is calculated, including:

[0059] For each fuzzy rule, fuzzy logic operations are performed using the fuzzy values ​​of the target feature. The rule weight is determined based on the activation degree of each fuzzy rule and the predefined evaluation rules. The rule weight of each fuzzy rule is then normalized to obtain the rule weight of the fuzzy rule corresponding to each target feature.

[0060] Specifically, firstly, for the fuzzy value of each target feature, fuzzy logic operations are performed with the corresponding fuzzy rules. For example, suppose there is a fuzzy rule that "if the engine temperature is medium and the speed is high, a malfunction may occur". In this case, this embodiment of the application needs to obtain the membership value of the engine temperature feature on the "medium" fuzzy set and the membership value of the speed on the "high" fuzzy set, and perform a logical "AND" operation.

[0061] Furthermore, the result of the above fuzzy logic operation is the activation degree of the fuzzy rule, representing the applicability of this rule in the current context. For example, if the obtained membership degree value of engine temperature is 0.7 and the membership degree value of engine speed is 0.8, then the activation degree of the above fuzzy rule is the minimum value of 0.7 and 0.8, which is 0.7.

[0062] Furthermore, a weight is assigned to each fuzzy rule according to predefined evaluation rules. These evaluation rules may be based on historical vehicle data, expert knowledge, or other relevant information. For example, based on past experience, higher engine temperatures and speeds generally indicate a higher probability of failure; therefore, the corresponding fuzzy rule might be assigned a higher weight.

[0063] Furthermore, to ensure that the sum of the weights of all fuzzy rules is 1, the weights of each fuzzy rule are normalized. For example, if there are three fuzzy rules with original weights of 0.5, 0.3, and 0.4, the normalized weights are 0.5 / 1.2 = 0.4167, 0.3 / 1.2 = 0.25, and 0.4 / 1.2 = 0.3333. Finally, by comprehensively considering the normalized weights of all fuzzy rules for each target feature, the final weight of that feature is determined, thus reflecting its importance in fault diagnosis.

[0064] In some embodiments, during the defuzzification process, an output member function is used to convert the fuzzy output into a specific numerical value or fault level. Based on the output of the adaptive neural fuzzy inference system, the health status of the target component is determined, including:

[0065] For each activated fuzzy rule, the fuzzy output value of the fuzzy rule is calculated based on the conclusion part of the fuzzy rule and the corresponding rule weight.

[0066] The fuzzy output is obtained by taking a weighted average of the fuzzy output value of each fuzzy rule and its corresponding rule weight.

[0067] Based on the output member function, the fuzzy output is mapped to a specific value or fault level using the defuzzification method.

[0068] Based on specific numerical values ​​or fault levels, the health status of target components is determined using preset fault judgment thresholds.

[0069] Specifically, for each activated fuzzy rule, a fuzzy output value is calculated based on the conclusion of the rule and its corresponding rule weight. For example, a fuzzy rule might be "If the engine temperature is high and the engine speed is high, then the risk of failure is high," and the fuzzy output value for "high" in the conclusion might be 0.8. If the weight of this rule is 0.6, then the weighted fuzzy output value is 0.8 x 0.6 = 0.48.

[0070] Furthermore, the fuzzy output values ​​of all activated fuzzy rules and their corresponding rule weights are weighted and averaged to obtain the total fuzzy output. For example, if there are three activated fuzzy rules with weighted fuzzy output values ​​of 0.48, 0.3, and 0.6, the total fuzzy output is (0.48 + 0.3 + 0.6) / 3 = 0.46.

[0071] Furthermore, based on a predefined output member function, a defuzzification method (e.g., the centroid method) is applied to map the fuzzy output value 0.46 to a specific numerical value or fault level. For example, this value can be between 0 and 1, where 1 represents the highest fault risk. Finally, based on the mapped specific numerical value or fault level, and combined with a preset fault judgment threshold, the health status of the target component is assessed. For example, the following judgment criteria can be set:

[0072] If the value is between 0 and 0.3, it is considered "healthy";

[0073] If the value is between 0.3 and 0.7, it is judged as "Caution";

[0074] If the value is between 0.7 and 1, it is considered "high risk".

[0075] Based on the defuzzification value of 0.46, the health status of the target component is determined to be "attention".

[0076] Using the methods described in the above embodiments, vehicles can assess the health status of their components in real time and provide timely warnings or maintenance for potential problems, thereby ensuring driving safety and improving the service life of the vehicle.

[0077] According to the technical solution provided in the embodiments of this application, this application collects and fuses multi-source vehicle data, such as the real-time status of target components, real-time alarms, external environment, map navigation, and weather data, to obtain more comprehensive vehicle operation information. Using these fused datasets, fault diagnosis can be performed more accurately. An adaptive neural fuzzy inference system is adopted, combining the adaptability of neural networks with the fuzzy reasoning ability to handle fuzziness, enabling the system to effectively determine faults even with uncertain or fuzzy data. As the training dataset accumulates, the system's judgment capability gradually improves, adapting to different working environments and conditions. For each target feature in the multi-source dataset, the system can automatically calculate its corresponding rule weight based on the degree of matching between each feature and the fuzzy rule. Thus, even with a large amount of input data, the system can quickly determine the health status of the target components, achieving efficient fault diagnosis. During the defuzzification process, the system uses output member functions to convert the fuzzy output into specific numerical values ​​or fault levels. This allows maintenance personnel or other users to intuitively understand the system's output results and make judgments or take appropriate measures accordingly. This solution fully considers various possible data anomalies and ensures the accuracy and reliability of the data input into the adaptive neurofuzzy inference system through preprocessing, data fusion, and feature extraction. Furthermore, after training on a training dataset, the system can continuously provide robust and reliable fault diagnosis services under various complex working conditions.

[0078] The following are embodiments of the apparatus described in this application, which can be used to execute the embodiments of the method described in this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the method described in this application.

[0079] Figure 2 This is a schematic diagram of the structure of a vehicle fault diagnosis device based on multi-source data fusion provided in an embodiment of this application. Figure 2 As shown, the vehicle fault diagnosis device based on multi-source data fusion includes:

[0080] The acquisition module 201 is configured to acquire real-time status data and real-time alarm data of the target components, acquire external environmental data using sensors installed on the vehicle, and obtain the vehicle's map navigation data.

[0081] The preprocessing module 202 is configured to perform preprocessing operations on the real-time status data, real-time alarm data, external environment data and map navigation data of the target components, and to fuse the preprocessed data to obtain a fused multi-source dataset.

[0082] The feature extraction module 203 is configured to extract features from a multi-source dataset to obtain target features for vehicle fault detection, and input the target features into a predetermined adaptive neurofuzzy inference system, wherein the adaptive neurofuzzy inference system includes a variety of fuzzy rules defined based on expert knowledge.

[0083] The fuzzy processing module 204 is configured to use an adaptive neural fuzzy inference system to map target features to different fuzzy sets and use input member functions to fuzzify the target features;

[0084] The calculation module 205 is configured to calculate the rule weight corresponding to each target feature for each fuzzy rule in the fuzzy set. The rule weight is used to characterize the fuzzy intensity corresponding to the target feature.

[0085] Training module 206 is configured to train the adaptive neurofuzzy inference system using a neural network learning algorithm and a training dataset in order to adjust the member functions and rule weights in the adaptive neurofuzzy inference system;

[0086] The output module 207 is configured to convert the fuzzy output into a specific value or fault level using the output member function during the defuzzification process, and to determine the health status of the target component based on the output of the adaptive neural fuzzy inference system.

[0087] In some embodiments, Figure 2 The data acquisition module 201 utilizes the status monitoring system installed on the vehicle to monitor the real-time status data generated by the target components. When an abnormal status or parameter exceeds the predetermined range is detected, real-time alarm data is generated. The environmental sensors on the vehicle collect the external environmental data of the vehicle in real time, obtain the vehicle's map navigation data from the vehicle's navigation system, and obtain the weather data of the expected driving route from the predetermined application.

[0088] In some embodiments, Figure 2 The preprocessing module 202 takes the real-time status data, real-time alarm data, external environment data, map navigation data, and weather data of the target components as raw data; performs data cleaning on the raw data to detect and delete missing, duplicate, or erroneous values; performs scaling processing on different types of raw data to unify raw data of different dimensions; and uses an autoencoder model to perform anomaly detection on the raw data to identify abnormal data.

[0089] In some embodiments, Figure 2The preprocessing module 202 weights the original data according to the fusion weights corresponding to the original data in different dimensions to obtain a weighted multi-source dataset; or, it performs feature fusion on the original data in different dimensions and merges multiple original data to obtain a fused multi-source dataset; or, it classifies the original data and performs category fusion on the classified original data according to the classification results to obtain a fused multi-source dataset.

[0090] In some embodiments, Figure 2 The fuzzy processing module 204 defines corresponding fuzzy variables based on the target features, and defines a fuzzy set for each fuzzy variable. It selects a member function to describe each fuzzy set, so that each target feature obtains a membership value on the corresponding fuzzy set. For each target feature, it uses the selected input member function to calculate its membership value on each fuzzy set, so as to convert each target feature into a fuzzy value. The calculated fuzzy value is stored in the fuzzy matrix for fuzzy inference and defuzzification operations.

[0091] In some embodiments, Figure 2 For each fuzzy rule, the calculation module 205 performs fuzzy logic operations using the fuzzy values ​​of the target features, determines the rule weight based on the activation degree of each fuzzy rule and the predefined evaluation rules, and normalizes the rule weight of each fuzzy rule to obtain the rule weight of the fuzzy rule corresponding to each target feature.

[0092] In some embodiments, Figure 2 For each activated fuzzy rule, the output module 207 calculates the fuzzy output value of the fuzzy rule based on the conclusion part of the fuzzy rule and the corresponding rule weight; it performs a weighted average based on the fuzzy output value of each fuzzy rule and the corresponding rule weight to obtain the fuzzy output; based on the output member function, it uses a defuzzification method to map the fuzzy output to a specific value or fault level; and based on the specific value or fault level, it uses a preset fault judgment threshold to determine the health status of the target component.

[0093] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0094] Figure 3 This is a schematic diagram of the structure of the electronic device 3 provided in an embodiment of this application. For example... Figure 3As shown, the electronic device 3 of this embodiment includes a processor 301, a memory 302, and a computer program 303 stored in the memory 302 and executable on the processor 301. When the processor 301 executes the computer program 303, it implements the steps in the various method embodiments described above. Alternatively, when the processor 301 executes the computer program 303, it implements the functions of each module / unit in the various device embodiments described above.

[0095] For example, computer program 303 may be divided into one or more modules / units, which are stored in memory 302 and executed by processor 301 to complete this application. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of computer program 303 in electronic device 3.

[0096] Electronic device 3 can be a desktop computer, laptop, handheld computer, cloud server, or other electronic device. Electronic device 3 may include, but is not limited to, processor 301 and memory 302. Those skilled in the art will understand that... Figure 3 This is merely an example of electronic device 3 and does not constitute a limitation on electronic device 3. It may include more or fewer components than shown, or combine certain components, or different components. For example, electronic device may also include input / output devices, network access devices, buses, etc.

[0097] Processor 301 can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0098] The memory 302 can be an internal storage unit of the electronic device 3, such as a hard disk or RAM. The memory 302 can also be an external storage device of the electronic device 3, such as a plug-in hard disk, Smart Media Card (SMC), Secure Digital (SD) card, or Flash Card. Furthermore, the memory 302 can include both internal and external storage units of the electronic device 3. The memory 302 is used to store computer programs and other programs and data required by the electronic device. The memory 302 can also be used to temporarily store data that has been output or will be output.

[0099] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0100] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0101] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0102] In the embodiments provided in this application, it should be understood that the disclosed apparatus / computer devices and methods can be implemented in other ways. For example, the apparatus / computer device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. Multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, and the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0103] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0104] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0105] If integrated modules / units are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program may include computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. Computer-readable media may include: any entity or device capable of carrying computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc.

[0106] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A vehicle fault diagnosis method based on multi-source data fusion, characterized in that, include: Real-time status data and real-time alarm data of target components are collected, external environmental data are collected using sensors installed on the vehicle, and map navigation data and weather data along the expected driving route of the vehicle are obtained. Preprocessing operations are performed on the real-time status data of the target component, the real-time alarm data, the external environment data, and the map navigation data. The preprocessed data are then fused to obtain a fused multi-source dataset. The fusion methods include one of the following: weighted fusion based on the fusion weights corresponding to the original data of different dimensions; feature fusion of the original data of different dimensions and mutual fusion of multiple original data; classification of the original data and category fusion of the classified original data; Feature extraction is performed on the multi-source dataset to obtain target features for vehicle fault detection. The target features are then input into a predetermined adaptive neurofuzzy inference system, wherein the adaptive neurofuzzy inference system includes multiple fuzzy rules defined based on expert knowledge. Based on the target features, fuzzy variables and fuzzy sets are defined. The adaptive neural fuzzy inference system is used to map the target features to different fuzzy sets. The target features are then fuzzified using input member functions so that each target feature is converted into a fuzzy value and the fuzzy value is stored in a fuzzy matrix. For each fuzzy rule in the fuzzy set, the rule weight corresponding to each target feature is calculated based on the activation degree of each fuzzy rule and a predefined evaluation rule. The rule weight is used to characterize the fuzziness intensity corresponding to the target feature. The adaptive neurofuzzy inference system is trained using a neural network learning algorithm and a training dataset in order to adjust the member functions and rule weights in the adaptive neurofuzzy inference system. During the defuzzification process, a weighted average is performed based on the fuzzy output value and the corresponding rule weight of each fuzzy rule to obtain the fuzzy output. The fuzzy output is then converted into a specific numerical value or fault level using the output member function. Based on the results output by the adaptive neural fuzzy inference system, the health status of the target component is determined.

2. The method according to claim 1, characterized in that, The process of collecting real-time status data and real-time alarm data of the target components, collecting external environmental data using sensors installed on the vehicle, and acquiring map navigation data and weather data along the expected driving route includes: The system uses a status monitoring system installed on the vehicle to monitor the real-time status data generated by the target components. When an abnormal status or parameter exceeds a predetermined range is detected, real-time alarm data is generated. The system also uses environmental sensors on the vehicle to collect real-time data on the external environment in which the vehicle is located, obtains map navigation data from the vehicle's navigation system, and obtains weather data for the expected driving route from a predetermined application.

3. The method according to claim 2, characterized in that, The preprocessing operation on the real-time status data of the target component, the real-time alarm data, the external environment data, and the map navigation data includes: The real-time status data of the target component, the real-time alarm data, the external environment data, the map navigation data, and the weather data are used as the raw data; The raw data is cleaned to detect and remove missing, duplicate, or erroneous values; different types of raw data are scaled to unify the raw data across different dimensions; and anomaly detection is performed on the raw data using an autoencoder model to identify anomalous data.

4. The method according to claim 1, characterized in that, The process involves defining fuzzy variables and fuzzy sets based on the target features, using the adaptive neural fuzzy inference system to map the target features to different fuzzy sets, and using input member functions to fuzzify the target features so as to convert each target feature into a fuzzy value, and storing the fuzzy values ​​in a fuzzy matrix, including: Based on the target features, define corresponding fuzzy variables, and define a fuzzy set for each fuzzy variable. Select a membership function to describe each fuzzy set, so that each target feature obtains a membership value on the corresponding fuzzy set. For each target feature, its membership value on each fuzzy set is calculated using a selected input membership function to convert each target feature into a fuzzy value. The calculated fuzzy value is stored in a fuzzy matrix for fuzzy inference and defuzzification operations.

5. The method according to claim 1, characterized in that, For each fuzzy rule in the fuzzy set, the rule weight corresponding to each target feature is calculated based on the activation degree of each fuzzy rule and a predefined evaluation rule, including: For each fuzzy rule, fuzzy logic operations are performed using the fuzzy value of the target feature. The rule weight is determined based on the activation degree of each fuzzy rule and a predefined evaluation rule. The rule weight of each fuzzy rule is then normalized to obtain the rule weight of the fuzzy rule corresponding to each target feature.

6. The method according to claim 1, characterized in that, In the defuzzification process, a weighted average is performed based on the fuzzy output value and corresponding rule weight of each fuzzy rule to obtain a fuzzy output. Using an output member function, the fuzzy output is converted into a specific numerical value or fault level. Based on the output of the adaptive neural fuzzy inference system, the health status of the target component is determined, including: For each activated fuzzy rule, the fuzzy output value of the fuzzy rule is calculated based on the conclusion part of the fuzzy rule and the corresponding rule weight. The fuzzy output is obtained by weighting the fuzzy output value of each fuzzy rule and its corresponding rule weight. Based on the output member function, the fuzzy output is mapped to a specific numerical value or fault level using a defuzzification method. Based on the specific numerical value or fault level, the health status of the target component is determined using a preset fault judgment threshold.

7. A vehicle fault diagnosis device based on multi-source data fusion, characterized in that, include: The data acquisition module is configured to collect real-time status data and real-time alarm data of the target components, collect external environmental data using sensors installed on the vehicle, and obtain map navigation data and weather data along the expected driving route of the vehicle. The preprocessing module is configured to perform preprocessing operations on the real-time status data of the target component, the real-time alarm data, the external environment data, and the map navigation data, and to fuse the preprocessed data to obtain a fused multi-source dataset. The fusion methods include one of the following: weighted fusion based on the fusion weights corresponding to the original data of different dimensions; feature fusion of the original data of different dimensions and mutual fusion of multiple original data; classification of the original data and category fusion of the classified original data; The feature extraction module is configured to extract features from the multi-source dataset to obtain target features for vehicle fault detection, and input the target features into a predetermined adaptive neurofuzzy inference system, wherein the adaptive neurofuzzy inference system includes multiple fuzzy rules defined based on expert knowledge. The fuzzy processing module is configured to define fuzzy variables and fuzzy sets based on the target features, use the adaptive neural fuzzy inference system to map the target features to different fuzzy sets, and use input member functions to fuzzify the target features so as to convert each target feature into a fuzzy value and store the fuzzy value in a fuzzy matrix; The calculation module is configured to calculate the rule weight corresponding to each target feature for each fuzzy rule in the fuzzy set based on the activation degree of each fuzzy rule and a predefined evaluation rule. The rule weight is used to characterize the fuzziness intensity corresponding to the target feature. The training module is configured to train the adaptive neurofuzzy inference system using a neural network learning algorithm and a training dataset in order to adjust the member functions and rule weights in the adaptive neurofuzzy inference system. The output module is configured to, during the defuzzification process, perform a weighted average of the fuzzy output value and the corresponding rule weight of each fuzzy rule to obtain a fuzzy output, use the output member function to convert the fuzzy output into a specific value or fault level, and determine the health status of the target component based on the results output by the adaptive neural fuzzy inference system.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 6.

Images