Data analysis system and method for vehicle electrical system fault diagnosis

Abstract

The invention provides a data analysis system and method for fault diagnosis of a vehicle electrical system. The method comprises the following steps: extracting multi-dimensional characteristic indexes representing the operation state of the electrical system from operation data of the electrical system under different working conditions; performing credibility evaluation on the corresponding multi-dimensional characteristic indexes under each working condition, determining credible evidences of the multi-dimensional characteristic indexes, and further determining credibility differences of the multi-dimensional characteristic indexes under different working conditions according to all the credible evidences; performing difference analysis on the data acquired under different working conditions according to the distribution correlation of the multi-dimensional characteristic indexes in the operation data to obtain the consistency difference of the multi-dimensional characteristic indexes under different working conditions; and according to the credibility difference and the consistency difference, performing feature fusion on all the collected operation data to generate fusion feature parameters capable of reflecting the operation state of the vehicle electrical system. According to the scheme of the invention, the state features of the electrical system can be fused through the credibility difference and the consistency difference of the features under multiple working conditions, so that the fault diagnosis robustness of the electrical system is improved.

Description

Technical Field

[0001] This application relates to the field of data analysis technology, and more specifically, to a data analysis system and method for fault diagnosis of vehicle electrical systems. Background Technology

[0002] In vehicle electrical systems, with the increasing complexity of electronic control units, intelligent sensors, and electronic loads, the system's operating status is affected by various factors. Monitoring a single operating condition or a single feature is insufficient to fully reflect the system's true operating status. State feature fusion, as a comprehensive analysis method, can effectively characterize the working status and performance change characteristics of each component of the electrical system by integrating operating information from different operating conditions and features. This provides more comprehensive and reliable data support for fault diagnosis, improves the accuracy and stability of diagnosis, and has significant engineering application value and significance.

[0003] However, existing vehicle electrical system fault diagnosis methods typically rely on feature analysis or simple feature overlay under a single operating condition, neglecting the reliability differences and correlation changes of features under multiple operating conditions. This leads to significant noise interference and unbalanced weight distribution during feature fusion, failing to accurately reflect the overall operating state of the system. In practical applications, the reliability of features collected by intelligent sensors varies significantly under different operating conditions, and the correlation structure between features may also fluctuate with changes in operating conditions. This makes directly fused features prone to deviating from the true system state, reducing the robustness and prediction accuracy of the diagnostic model. Therefore, how to achieve the fusion of electrical system state features by leveraging the reliability and consistency differences of features under multiple operating conditions, thereby improving the robustness of electrical system fault diagnosis, has become a challenge for the industry. Summary of the Invention

[0004] This application provides a data analysis system and method for fault diagnosis of vehicle electrical systems. It can achieve the fusion of electrical system state characteristics by the reliability and consistency differences of features under multiple operating conditions, thereby improving the robustness of electrical system fault diagnosis.

[0005] In a first aspect, this application provides a data analysis method for fault diagnosis of vehicle electrical systems, comprising the following steps: Intelligent sensors are used to collect operating data of the electrical system of the target vehicle under different working conditions. Based on the operational data, multi-dimensional feature indicators characterizing the operating status of the electrical system are extracted. These multi-dimensional feature indicators are used to reflect the working status and performance change characteristics of each component of the electrical system. The credibility of the multidimensional feature indicators corresponding to each working condition is evaluated, and their credible evidence is determined. Then, the credibility difference of the multidimensional feature indicators under different working conditions is determined by all the credible evidence. Based on the distribution correlation of multidimensional feature indicators in each operational data, a difference analysis was conducted on the data collected under different working conditions to obtain the consistency differences of multidimensional feature indicators under different working conditions. Based on the credibility difference and the consistency difference, all collected operational data are fused to generate fused feature parameters that can reflect the operating status of the target vehicle's electrical system. The fused feature parameters are then input into the fault diagnosis model to achieve intelligent diagnosis of the fault status of the vehicle's electrical system.

[0006] Preferably, based on the operational data, extracting multidimensional feature indicators characterizing the operating state of the electrical system specifically includes: Preprocess the operating data of the target vehicle's electrical system, including voltage, current, power, and output commands from the electronic control unit; Characteristic factors are extracted from the preprocessed operating data to characterize electrical load characteristics, power transmission stability, and electronic control response rate. The various characteristic factors are combined to form a multidimensional characteristic index, which is used to characterize the overall operating status of the target vehicle's electrical system.

[0007] Preferably, the credibility of the multi-dimensional characteristic indicators corresponding to each working condition is assessed, and the specific evidence for their credibility is determined to include: For multi-dimensional characteristic indicators under various working conditions, a characteristic confidence information set reflecting the fluctuation range, characteristic stability and measurement noise level of the indicators is constructed to quantify the confidence level of each characteristic indicator. Based on the aforementioned feature confidence information set, a multi-source feature credibility assessment model is established according to the DS evidence theory, which transforms confidence information from different feature indicator sources into corresponding evidence support. The evidence support of each feature indicator across different data sources is fused to obtain credible evidence for the corresponding feature indicator under the current working conditions.

[0008] Preferably, the determination of the credibility differences of multidimensional feature indicators under different working conditions based on all credible evidence specifically includes: A standard credibility matrix is ​​constructed based on all credible evidence to measure the differences in credible distributions between operating conditions. Using the standard credibility matrix, the credibility distribution of each feature index under different working conditions is cross-compared to calculate the evidence support bias and evidence concentration bias between working conditions. Based on the evidence support bias and the evidence set bias, a credibility difference mapping table is constructed to characterize the working condition sensitivity of the feature indicators. Then, the credibility difference of the multidimensional feature indicators under different working conditions is determined through the credibility difference mapping table.

[0009] Preferably, based on the distribution correlation of multidimensional characteristic indicators in each operational data, a difference analysis is performed on the data collected under different working conditions to obtain the consistency differences of multidimensional characteristic indicators under different working conditions, specifically including: Extract the distribution correlation between multidimensional feature indicators under various working conditions; Comparative analysis of all distribution correlation degrees was performed to extract the difference information of characteristic correlation structure between various working conditions; Based on the aforementioned difference information, the consistency differences of multidimensional feature indicators under different working conditions are determined.

[0010] Preferably, the process of fusing features from all collected operational data based on the reliability differences and consistency differences to generate fused feature parameters that reflect the operating status of the target vehicle's electrical system specifically includes: Construct a fusion weight matrix for the operating data under different working conditions based on the credibility difference and the consistency difference; The fusion weight matrix is ​​used to aggregate multi-dimensional feature indicators under different working conditions to generate a comprehensive feature vector. The comprehensive feature vector is used as input to generate fused feature parameters that can reflect the operating status of the target vehicle's electrical system.

[0011] Preferably, the operating data of the electrical system of the target vehicle under different working conditions is collected by the data acquisition unit on the vehicle.

[0012] Secondly, this application provides a data analysis system for fault diagnosis of vehicle electrical systems, comprising: The data acquisition module is used to collect operating data of the electrical system of the target vehicle under different working conditions through intelligent sensors; The feature processing module is used to extract multi-dimensional feature indicators that characterize the operating status of the electrical system based on the operating data. The multi-dimensional feature indicators are used to reflect the working status and performance change characteristics of each component of the electrical system. The feature processing module is also used to evaluate the credibility of the multi-dimensional feature indicators corresponding to each working condition and determine their credible evidence, and then determine the credibility difference of the multi-dimensional feature indicators under different working conditions based on all the credible evidence. The feature processing module is also used to perform difference analysis on the data collected under different working conditions based on the distribution correlation of multidimensional feature indicators in each operating data, and to obtain the consistency difference of multidimensional feature indicators under different working conditions. The feature fusion module is used to fuse all collected operational data based on the confidence difference and the consistency difference to generate fused feature parameters that can reflect the operating status of the target vehicle's electrical system. The fused feature parameters are then input into the fault diagnosis model to achieve intelligent diagnosis of the fault status of the vehicle's electrical system.

[0013] Thirdly, this application provides a computer device, the computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described data analysis method for diagnosing faults in vehicle electrical systems.

[0014] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described data analysis method for diagnosing faults in vehicle electrical systems.

[0015] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects: In this embodiment, intelligent sensors are used to collect operational data of the electrical system of a target vehicle under different operating conditions. Based on the operational data, multidimensional feature indicators characterizing the operating state of the electrical system are extracted. These multidimensional feature indicators reflect the operating state and performance change characteristics of each component of the electrical system. The credibility of the multidimensional feature indicators corresponding to each operating condition is evaluated, and credible evidence is determined. Then, the credibility difference of the multidimensional feature indicators under different operating conditions is determined by all credible evidence. Based on the distribution correlation of the multidimensional feature indicators in each operational data, the data collected under different operating conditions is analyzed for differences, and the consistency difference of the multidimensional feature indicators under different operating conditions is obtained. Based on the credibility difference and the consistency difference, all collected operational data are fused to generate fused feature parameters that can reflect the operating state of the target vehicle's electrical system. Then, the fused feature parameters are input into the fault diagnosis model to realize intelligent diagnosis of the fault state of the vehicle's electrical system.

[0016] Therefore, this application systematically processes multidimensional operating data collected from vehicle electrical systems under different operating conditions, realizing a complete process from feature extraction, reliability assessment, difference analysis to feature fusion, thus improving the overall accuracy and robustness of electrical system fault diagnosis. First, the reliability of multidimensional feature indicators corresponding to each operating condition is assessed, and credible evidence is determined. Then, the reliability differences of multidimensional feature indicators under different operating conditions are determined by all credible evidence. Through these reliability differences, unreliable features significantly affected by sensor noise, measurement errors, and transient fluctuations under different operating conditions can be clearly distinguished from highly reliable features, effectively reducing the interference of abnormal data on diagnostic results and ensuring that the contributions of each feature during the fusion process are reasonably weighted, enhancing the ability of comprehensive features to depict the true state of the system. Second, based on the distribution correlation of multidimensional feature indicators in each operating data, difference analysis is performed on the data collected under different operating conditions to obtain the consistency differences of multidimensional feature indicators under different operating conditions. Through these consistency differences... This approach can deeply reveal the dynamic correlation structure and sensitivity of features across operating conditions, identify features that are inconsistent under specific operating conditions and easily affected by load changes or external interference, and provide a quantitative basis for appropriate adjustment of sensitive features during the fusion process, thereby effectively avoiding diagnostic biases caused by differences in operating conditions. Then, based on the reliability differences and consistency differences, all collected operating data are fused to generate fused feature parameters that reflect the operating status of the target vehicle's electrical system. Through feature fusion, reliable information across operating conditions and features can be integrated into a unified representation vector, which not only improves the completeness and stability of feature expression but also enhances the robustness of the fault diagnosis model in the face of complex and dynamic operating conditions, ensuring that the diagnostic results can accurately reflect the true state of the system, while suppressing the impact of occasional anomalies or sudden events on the overall diagnosis. In summary, the solution of this application can achieve the fusion of electrical system state features through the reliability differences and consistency differences of features under multiple operating conditions, thereby improving the robustness of electrical system fault diagnosis. Attached Figure Description

[0017] Figure 1 This is an exemplary flowchart of a data analysis method for fault diagnosis of a vehicle electrical system, according to some embodiments of this application; Figure 2 This is a schematic diagram of the process for determining confidence differences according to some embodiments of this application; Figure 3 This is a flowchart illustrating the process of determining consistency differences according to some embodiments of this application; Figure 4 This is a schematic diagram of the structure of a data analysis system for fault diagnosis of vehicle electrical systems according to some embodiments of this application; Figure 5This is a schematic diagram of the structure of a computer device for implementing a data analysis method for fault diagnosis of a vehicle electrical system, according to some embodiments of this application. Detailed Implementation

[0018] To better understand the technical solution of this application, the technical solution of this application will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] refer to Figure 1 The figure is an exemplary flowchart of a data analysis method for fault diagnosis of a vehicle electrical system according to some embodiments of this application. The data analysis method for fault diagnosis of a vehicle electrical system mainly includes the following steps: In step 101, the operating data of the electrical system of the target vehicle under different working conditions are collected by intelligent sensors.

[0020] In some embodiments, the electrical system operating data of the target vehicle under different operating conditions is collected by a data acquisition unit mounted on the vehicle. The data acquisition unit includes a vehicle controller interface module, a sensor module, and a data recording device. The vehicle controller interface module is used to acquire the output command signals and control status of the electronic control unit. The sensor module is used to collect voltage, current, power, and temperature signals. The data recording device is used to synchronously store the signals and timestamp them. Data acquisition is carried out under various operating conditions of the vehicle, including engine start / stop, acceleration and deceleration, constant speed driving, and braking. The collected data is imported into local storage or a remote server to provide a data foundation for subsequent multi-dimensional feature index extraction and fault diagnosis. It should be further noted that the operating data specifically includes the voltage, current, power, temperature, and output command signals of the vehicle's electrical system and electronic control unit.

[0021] In step 102, based on the operating data, multidimensional feature indicators characterizing the operating status of the electrical system are extracted. These multidimensional feature indicators are used to reflect the working status and performance change characteristics of each component of the electrical system.

[0022] In some embodiments, extracting multidimensional feature indicators characterizing the operating state of an electrical system based on the operational data can be achieved through the following steps: Preprocess the operating data of the target vehicle's electrical system, including voltage, current, power, and output commands from the electronic control unit; Characteristic factors are extracted from the preprocessed operating data to characterize electrical load characteristics, power transmission stability, and electronic control response rate. The various characteristic factors are combined to form a multidimensional characteristic index, which is used to characterize the overall operating status of the target vehicle's electrical system.

[0023] It should be noted that the electronic control response rate in this application describes the response speed performance of the electronic control system to command signals; the multidimensional characteristic index is a set of multiple characteristic factors used to comprehensively characterize the operating state of the vehicle's electrical system, and is used to reflect the working state and performance change characteristics of each component of the electrical system.

[0024] In practical implementation, firstly, the operating data of the target vehicle's electrical system is preprocessed. Specific operations include: removing outliers from the collected voltage, current, power, and electronic control unit (ECU) output command signals by using threshold judgment or statistical analysis to eliminate abrupt changes exceeding reasonable ranges; performing low-pass filtering on current and voltage signals to remove high-frequency noise, using existing finite impulse response (FIR) filters or moving average methods; and smoothing power and ECU command signals to eliminate instantaneous fluctuation interference. Then, characteristic factors characterizing the electrical system's operating state are extracted from the preprocessed data. Electrical load characteristics can be characterized by calculating the instantaneous rate of change, mean, and peak value of current and power to represent load fluctuations. Combined with existing power monitoring technology, local change trends can be calculated using a sliding window. Power transmission stability can be obtained by analyzing the standard deviation, variance, and fluctuation amplitude of voltage and power signals to reflect the system's energy transmission stability under different operating conditions. The ECU response rate can be determined by comparing the delay of the ECU command signal with the actual response signal. The rise time and slope of the response curve are used to characterize the system, which can be achieved using existing response time analysis methods. It should be noted that the extraction of each type of feature factor is based on well-known data analysis and signal processing techniques, and can quantify different aspects of system operation. Further details are omitted here. Finally, the different feature factors are arranged in a fixed order to form a vector or matrix, and the indicators such as electrical load characteristics, power transmission stability, and electronic control response rate are integrated into a single multidimensional feature vector to simultaneously characterize the comprehensive operating state of the target vehicle's electrical system. This process can be achieved through standard vectorization methods, matrix concatenation, or multidimensional array combination. This is only an example and is not intended to limit the specific scope of the invention.

[0025] In step 103, the credibility of the multidimensional feature indicators corresponding to each working condition is evaluated, and their credibility evidence is determined. Then, the credibility difference of the multidimensional feature indicators under different working conditions is determined by all the credibility evidence.

[0026] In some embodiments, the credibility assessment of multidimensional feature indicators corresponding to various working conditions and the determination of their credible evidence can be achieved through the following steps: For multi-dimensional characteristic indicators under various working conditions, a characteristic confidence information set reflecting the fluctuation range, characteristic stability and measurement noise level of the indicators is constructed to quantify the confidence level of each characteristic indicator. Based on the aforementioned feature confidence information set, a multi-source feature credibility assessment model is established according to the DS evidence theory, which transforms confidence information from different feature indicator sources into corresponding evidence support. The evidence support of each feature indicator across different data sources is fused to obtain credible evidence for the corresponding feature indicator under the current working conditions.

[0027] It should be noted that the feature confidence information set in this application refers to the confidence data set used to comprehensively characterize the volatility, stability and measurement noise level of each feature indicator; the evidence support degree refers to the quantitative value reflecting the credibility of a feature indicator under a specific data source; and credible evidence is a comprehensive evaluation result used to quantify the reliability of each feature indicator under the current working conditions and support fault diagnosis.

[0028] In practice, firstly, the time-series data of each characteristic indicator under the current operating condition is sampled and smoothed to eliminate the impact of instantaneous fluctuations on the results. Then, the fluctuation range, characteristic stability, and measurement noise level of the characteristic indicator are calculated. The fluctuation range is obtained by the difference between the maximum and minimum values ​​of the characteristic indicator, which reflects the degree of data dispersion. The characteristic stability is obtained by calculating the standard deviation or coefficient of variation using the sliding window method, which reflects the stability of the characteristic over time. The measurement noise level is calculated based on the signal-to-noise ratio using sensor calibration data or repeated measurement results, which measures the magnitude of measurement error. Furthermore, the three types of parameters are normalized respectively, and then weighted and fused according to preset weights (e.g., fluctuation range weight is 0.4, stability weight is 0.4, and measurement noise weight is 0.2) to obtain the confidence value of each characteristic indicator under the current operating condition. Finally, the confidence values ​​corresponding to each characteristic indicator are arranged and combined according to the characteristic dimension to form a complete set of characteristic confidence information. Secondly, the feature confidence information from different sensing channels such as voltage, current, and spectral signals is standardized. The confidence level of each channel is normalized by interval and probability mapped to obtain the basic probability allocation function representing the credible contribution of each feature to the system's operating state. Then, based on the statistical independence and information complementarity between feature sources, a feature source conflict degree matrix is ​​constructed. By calculating the conflict coefficient between different feature confidence information, a suppression coefficient is applied to the feature source confidence information with high conflict degree to achieve weighted correction of multi-source confidence information. Furthermore, based on the DS (Dempster-Shafer) evidence synthesis rule, the corrected multi-source confidence information is fused step by step to calculate the comprehensive evidence support degree corresponding to each potential fault state, forming the confidence distribution of the target system under each fault hypothesis. During the fusion process, an evidence conflict normalization processing strategy is introduced to ensure that the overall evaluation result still has convergence and consistency when there are partial contradictions in the multi-source features. Finally, the fused comprehensive evidence support degree is used as the preliminary credible evidence output of the feature index under the current operating conditions. Then, the preliminary evidence support levels of the same feature index in different sensors or acquisition channels are uniformly normalized to eliminate differences in sampling frequency, noise interference, and measurement accuracy among data sources. Based on the historical stability and measurement error level of each data source, corresponding weight coefficients are assigned, giving data sources with higher weights a stronger influence in subsequent fusion. Furthermore, a weighted fusion rule based on DS evidence theory is adopted to perform pairwise synthesis of the preliminary evidence support levels from different sources. After each synthesis, conflict coefficients are calculated to adjust contradictory evidence, ensuring the confidence consistency of the final fusion result. Finally, the fused comprehensive support level is output as credible evidence of the feature index under the current working conditions.

[0029] It should be noted that the multi-source feature credibility assessment model in this application is used to quantify and fuse multi-dimensional feature information from different sensors or acquisition channels. Its specific implementation includes the following steps: First, the confidence information of each feature index under different data sources is transformed into a basic probability allocation function to characterize the credible contribution of each feature to the operating state of the vehicle's electrical system; second, based on the statistical independence and information complementarity between feature sources, a feature source conflict degree matrix is ​​constructed. By calculating the conflict coefficient between the confidence information of each feature, a weighted correction is applied to feature sources with high conflict degrees to suppress the influence of unreliable data on the fusion result; then… Based on the DS evidence synthesis rules, the modified multi-source confidence information is fused step by step to calculate the comprehensive evidence support degree corresponding to each potential fault state, forming the confidence distribution of the target system under each fault hypothesis. During the fusion process, an evidence conflict normalization processing strategy is adopted to ensure that the overall evaluation results still have convergence and consistency when there are partial contradictions in the multi-source features. Finally, the fused comprehensive evidence support degree is used as the credible evidence output of each feature index under the current working condition, thereby realizing the reliability quantification of multi-channel feature information and providing robust and credible feature basis for subsequent vehicle electrical system fault diagnosis.

[0030] In some embodiments, the credibility differences of multidimensional feature indicators under different operating conditions are determined by all credible evidence, with reference to... Figure 2 As shown in the figure, this is a schematic flowchart of determining the confidence difference in some embodiments of this application. The determination of the confidence difference can be achieved by the following steps: The following steps can be used to achieve this: In step 1031, a standard confidence matrix is ​​constructed based on all credible evidence to measure the differences in confidence distribution among operating conditions; In step 1032, the credibility distribution of each feature index under different working conditions is cross-compared using the standard credibility matrix to calculate the evidence support bias and evidence concentration bias between working conditions. In step 1033, a credibility difference mapping table is constructed to characterize the working condition sensitivity of the feature indicators based on the evidence support bias and the evidence set bias. Then, the credibility difference of the multidimensional feature indicators under different working conditions is determined through the credibility difference mapping table.

[0031] It should be noted that the standard credibility matrix in this application is a matrix used to describe the differences in the distribution of credible evidence for each feature indicator under different working conditions; the evidence support bias is an indicator that measures the change in the credibility of the same feature under different working conditions; the evidence concentration bias is an indicator that measures the degree of dispersion of the credibility distribution of the same feature under different working conditions; the credibility difference mapping table is a table that reflects the changes in the sensitivity and credibility of each feature indicator under different working conditions; and the credibility difference is an indicator that measures the change in the reliability of multidimensional feature indicators under different working conditions.

[0032] In specific implementation, firstly, the credible evidence of each feature indicator under different working conditions is arranged according to the feature dimension and the working condition dimension to form a standard credibility matrix. Each row of the matrix represents the credible evidence value of a certain feature indicator under each working condition, and each column represents the credibility distribution of all feature indicators under a certain working condition. This matrix can be constructed using known matrix construction methods, such as using a two-dimensional array for storage or spreadsheet software for data organization; this is merely an example and not intended to limit the specific scope of the invention. Secondly, based on the constructed standard credibility matrix, the credibility distribution of each feature indicator under different working conditions is cross-compared. By calculating statistical measures such as the difference in mean, variance, or standard deviation between each working condition, evidence support bias and evidence concentration bias are obtained. Evidence support bias reflects the shift in credibility of the same feature under different working conditions, while evidence concentration bias reflects the degree of dispersion or concentration of credibility of the same feature under different working conditions. The calculation of these statistical measures can be completed using commonly known algorithms such as mean calculation, variance formula, or standard deviation formula. The examples provided are merely illustrative and not intended to limit the scope of the invention. Next, the evidence support bias and evidence settling bias of each feature indicator under different operating conditions are normalized and mapped to a unified quantification range to eliminate differences in the dimensions and amplitudes of different indicators. This step can be achieved using common linear normalization or min-max normalization methods. Then, based on the normalized bias data, the comprehensive sensitivity value of each feature indicator under different operating conditions is calculated using weighted scoring or comprehensive indicators. The weighting coefficients can be set according to feature importance or empirical rules. This step can be completed using known linear weighted calculations or matrix operations. Further, the comprehensive sensitivity values ​​are divided into low-sensitivity, medium-sensitivity, and high-sensitivity levels according to preset levels, forming cell records in a mapping table. Rows in the mapping table correspond to feature indicators, columns correspond to different operating conditions, and cell content records the sensitivity level or comprehensive deviation value of the feature under that operating condition. Finally, by reading the sensitivity values ​​or levels of each feature indicator in the mapping table under each operating condition, the credibility differences of the multidimensional feature indicators under different operating conditions can be determined.

[0033] It should be noted that in this embodiment, by evaluating the credibility of multi-dimensional feature indicators under various working conditions and fusing multi-source features to generate credible evidence based on DS evidence theory, a standard credibility matrix and a credibility difference mapping table are constructed based on all credible evidence. This determines the credibility differences of multi-dimensional feature indicators under different working conditions. Compared with the prior art, this embodiment can solve the problems of large noise interference, high feature heterogeneity, and difficulty in quantifying working condition sensitivity caused by relying on only a single feature or directly statistical feature values ​​in traditional methods. The technical effect of this embodiment is that it can accurately characterize the reliability changes and working condition sensitivity of each feature indicator under different working conditions, providing a quantitative basis for feature fusion weighting and priority adjustment, thereby significantly improving the accuracy, stability, and robustness of vehicle electrical system fault diagnosis, demonstrating the inventiveness of the present invention.

[0034] In step 104, the data collected under different working conditions are analyzed for differences based on the distribution correlation of multidimensional feature indicators in each operating data, so as to obtain the consistency differences of multidimensional feature indicators under different working conditions.

[0035] In some embodiments, difference analysis is performed on data collected under different working conditions based on the distribution correlation of multidimensional feature indicators in each operating data, to obtain the consistency differences of multidimensional feature indicators under different working conditions, and to refer to... Figure 3 As shown in the figure, this is a flowchart illustrating the process of determining consistency differences in some embodiments of this application. The determination of consistency differences can be achieved through the following steps: In step 1041, the distribution correlation degree between multidimensional feature indicators under each working condition is extracted; In step 1042, a comparative analysis of all distribution correlation degrees is performed to extract the difference information of characteristic correlation structures between each working condition; In step 1043, the consistency differences of multidimensional feature indicators under different working conditions are determined based on the difference information.

[0036] It should be noted that the distribution correlation degree in this application is an indicator that measures the statistical correlation and dependency between multidimensional feature indicators; the consistency difference is an indicator that measures the stability change of the correlation structure of the same feature under different working conditions.

[0037] In practical implementation, firstly, the distribution correlation of multidimensional feature indicators under each working condition is extracted. For each working condition, including voltage, current, power, and other relevant feature factors, the data relationship between any two feature indicators is considered. When two feature indicators are linearly correlated, the Pearson correlation coefficient can be used to calculate their correlation. The calculation formula is r equal to the sum of the products of the deviations of each sample from the mean divided by the square root of the sum of the squares of the deviations from the mean, where x and y represent the datasets of the two features, and the mean of x and the mean of y represent the mean of the corresponding data. This calculation can be performed using the well-known ExcelCORREL function or the scipy library in Python. The `pearsonr` function is used for implementation. When the feature relationship is non-linear or the data is ordered, the Spearman rank correlation coefficient is used to calculate the correlation. First, the datasets of the two features are sorted and assigned ranks. Then, the correlation is calculated using a similar logic to the Pearson coefficient. This calculation can use the `spearmanr` function from the `scipy` library in Python. Using the above method, a feature correlation matrix is ​​constructed for each work condition. The matrix dimension is the product of the total number of feature indicators and the matrix rows and columns correspond to each feature indicator. The diagonal positions are filled with 1 to indicate that they are fully correlated. The remaining positions are filled with the calculated correlation values, thus completing the single work condition distribution correlation. The process involves two steps: first, extracting the degree of correlation; second, comparing and analyzing the distribution correlation of all working conditions to extract differences. The correlation matrices of all working conditions are paired, for example, starting and constant speed, acceleration and braking. For each pair of matrices, the known element difference method is used to align the matrix positions one by one, calculate the absolute difference of corresponding elements, and then calculate the mean and standard deviation of all differences. The threshold is set to the mean plus twice the standard deviation. This setting is a known statistical method that can effectively eliminate random fluctuations. If the difference at a certain position exceeds the threshold, the feature pair corresponding to that position is marked as a feature pair showing differences in the correlation structure between working conditions. This process is repeated for all working condition pairs. The process involves comparing and collecting all pairs of differing features to form a set of information on the differences in the correlation structure between operating conditions, i.e., the difference information on the feature correlation structure between each operating condition. Then, based on the difference information, consistency differences are determined. A counting statistical method can be used to count the total number of times each feature indicator participates in a pair of differing features. For example, if feature 1 and feature 2, and feature 1 and feature 3 are both pairs of differing features, then feature 1 has two differences. According to the statistical results, features with zero to one difference are defined as low consistency differences, indicating that the correlation structure of this feature is stable under different operating conditions; features with two to three differences are defined as medium consistency differences, indicating that the feature correlation structure has moderate fluctuations.Features with four or more differences are defined as having high consistency differences, indicating significant fluctuations in the feature association structure. Finally, the number of differences and corresponding difference levels for each feature indicator are compiled, and they can be sorted from high to low according to the difference level. This clearly reveals the consistency differences of multi-dimensional feature indicators under different working conditions, guiding subsequent feature fusion and weight allocation, and achieving robust representation of the operating state of multi-dimensional features under different working conditions. It should be noted that the standardization processing, correlation calculation, matrix comparison, difference statistics, and consistency level classification described in this embodiment can all be implemented using existing statistical analysis, matrix operations, and information theory methods, and are fully disclosed; no specific algorithm is limited.

[0038] In step 105, all collected operational data are fused based on the confidence difference and the consistency difference to generate fused feature parameters that can reflect the operating status of the target vehicle's electrical system. The fused feature parameters are then input into the fault diagnosis model to achieve intelligent diagnosis of the fault status of the vehicle's electrical system.

[0039] In some embodiments, the following steps can be used to perform feature fusion on all collected operational data based on the confidence difference and the consistency difference to generate fused feature parameters that can reflect the operating status of the target vehicle's electrical system: Construct a fusion weight matrix for the operating data under different working conditions based on the credibility difference and the consistency difference; The fusion weight matrix is ​​used to aggregate multi-dimensional feature indicators under different working conditions to generate a comprehensive feature vector. The comprehensive feature vector is used as input to generate fused feature parameters that can reflect the operating status of the target vehicle's electrical system.

[0040] It should be noted that the fusion weight matrix in this application refers to the matrix used to represent the reliability weight distribution of each characteristic index under different operating conditions; the fusion characteristic parameter refers to the comprehensive characteristic vector obtained by weighted integration of multi-dimensional characteristic indices under various operating conditions, which can comprehensively characterize the system's operating status.

[0041] In practice, to achieve the fusion processing of multi-dimensional feature indicators, the previously obtained reliability difference values ​​and consistency difference values ​​are first preprocessed. The two types of difference values ​​are mapped to the range of zero to one using the minimum-maximum normalization method to eliminate the influence of dimensional differences on subsequent weight calculations. Then, according to the well-known linear weighting principle, the normalized reliability difference values ​​and consistency difference values ​​are combined in equal proportions to calculate the weight of each feature indicator under each working condition. Specifically, the weight value is obtained by subtracting the normalized difference value from 1, so that features with smaller differences receive higher weights, thereby reflecting their reliability under different working conditions. Furthermore, all the calculated weights are arranged according to rows corresponding to working conditions and columns corresponding to feature indicators to form a fusion weight matrix. Then, the multi-dimensional feature indicators under each working condition are organized into a feature matrix, where the rows of the matrix represent each working condition and the columns represent each feature indicator. The corresponding elements are multiplied one by one with the elements of the fusion weight matrix to obtain the weighted feature matrix, so that the reliability of each feature indicator is adjusted according to its weight. Next, the weighted feature matrix is ​​summed column-wise to generate a one-dimensional comprehensive feature vector with a length equal to the number of feature indicators. Each element corresponds to the weighted comprehensive value of a feature indicator under all operating conditions. Finally, the comprehensive feature vector is standardized using the z-score standardization method, which subtracts the mean from each element in the vector and divides it by the standard deviation to eliminate the differences in the numerical range between different feature indicators, thus obtaining the final fused feature parameters. These fused feature parameters can objectively and comprehensively characterize the operating status of the target vehicle's electrical system and can be directly input into the subsequent fault diagnosis model to achieve intelligent diagnosis of electrical system fault states.

[0042] In some embodiments, fused feature parameters are input into a fault diagnosis model to achieve intelligent diagnosis of fault states in vehicle electrical systems. Specifically, the fault diagnosis model is used to automatically identify the operating mode, potential anomalies, and fault categories of the target vehicle electrical system based on the cross-condition comprehensive state information contained in the fused feature parameters. The fault diagnosis model can be constructed by a deep neural network model, a time-series discriminant model, or a graph-based association reasoning model. It achieves modeling of the operating rules of the vehicle electrical system and identification of abnormal behaviors by performing high-dimensional feature mapping, hidden-layer state association learning, and output-layer pattern classification on the fused feature parameters. This is only an example and is not intended to limit the specific scope of the invention. Since the fused feature parameters have undergone cross-condition feature calibration and unification processing based on confidence and consistency differences before input, the fault diagnosis model can effectively reduce the impact of noise data and operating condition offsets on feature discrimination during inference, thereby improving the accuracy and stability of fault determination results and enhancing the adaptability of the diagnostic system in complex operating environments.

[0043] On the other hand, in some embodiments, this application provides a data analysis system for fault diagnosis of vehicle electrical systems, with reference to... Figure 4 The figure is a schematic diagram of the structure of a data analysis system for vehicle electrical system fault diagnosis according to some embodiments of this application. The data analysis system for vehicle electrical system fault diagnosis includes: a data acquisition module 401, a feature processing module 402, and a feature fusion module 403, which are described below: Data acquisition module 401, in this application, is mainly used to collect the operating data of the electrical system of the target vehicle under different working conditions through intelligent sensors; Feature processing module 402, in this application, is used to extract multi-dimensional feature indicators that characterize the operating status of the electrical system based on the operating data. The multi-dimensional feature indicators are used to reflect the working status and performance change characteristics of each component of the electrical system. In this application, the feature processing module 402 is also used to evaluate the credibility of the multi-dimensional feature indicators corresponding to each working condition and determine their credible evidence, and then determine the credibility difference of the multi-dimensional feature indicators under different working conditions based on all the credible evidence. In this application, the feature processing module 402 is also used to perform difference analysis on the data collected under different working conditions based on the distribution correlation of multidimensional feature indicators in each operating data, and to obtain the consistency difference of multidimensional feature indicators under different working conditions. Feature fusion module 403: In this application, feature fusion module 403 is mainly used to perform feature fusion on all collected operating data based on the credibility difference and the consistency difference, generate fused feature parameters that can reflect the operating status of the target vehicle electrical system, and then input the fused feature parameters into the fault diagnosis model to realize intelligent diagnosis of the fault status of the vehicle electrical system.

[0044] In addition, this application also provides a computer device, the computer device including a memory and a processor, the memory storing code, and the processor being configured to acquire the code and execute the above-described data analysis method for fault diagnosis of vehicle electrical systems.

[0045] In some embodiments, reference Figure 5 The figure is a schematic diagram of the structure of a computer device implementing a data analysis method for fault diagnosis of a vehicle electrical system, according to some embodiments of this application. The data analysis method for fault diagnosis of a vehicle electrical system in the above embodiments can be... Figure 5 The computer device shown is used to implement this, and the computer device 500 includes at least one processor 501, a communication bus 502, a memory 503, and at least one communication interface 504.

[0046] Processor 501 can be a general-purpose central processing unit (CPU) or an application-specific integrated circuit (ASIC).

[0047] The communication bus 502 can be used to transmit information between the aforementioned components.

[0048] Memory 503 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital versatile optical discs, Blu-ray discs, etc.), magnetic disks or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. Memory 503 may exist independently and be connected to processor 501 via communication bus 502. Memory 503 may also be integrated with processor 501.

[0049] The memory 503 stores program code for executing the solution of this application, and its execution is controlled by the processor 501. The processor 501 executes the program code stored in the memory 503. The program code may include one or more software modules. The data analysis method for vehicle electrical system fault diagnosis in the above embodiments can be implemented by the processor 501 and one or more software modules in the program code in the memory 503.

[0050] Communication interface 504 uses any transceiver-like device to communicate with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area networks (WLAN), etc.

[0051] In a specific implementation, as one example, a computer device may include multiple processors, each of which may be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. Here, a processor may refer to one or more devices, circuits, and / or processing cores used to process data (e.g., computer program instructions).

[0052] The aforementioned computer device can be a general-purpose computer device or a special-purpose computer device. In specific implementations, the computer device can be a desktop computer, a portable computer, a network server, a handheld digital assistant (PDA), a mobile phone, a tablet computer, a wireless terminal device, a communication device, or an embedded device. This application does not limit the type of computer device.

[0053] In addition, this application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described data analysis method for fault diagnosis of vehicle electrical systems.

[0054] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0055] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A data analysis method for fault diagnosis of vehicle electrical systems, characterized in that, Includes the following steps: Intelligent sensors are used to collect operating data of the electrical system of the target vehicle under different working conditions. Based on the operational data, multi-dimensional feature indicators characterizing the operating status of the electrical system are extracted. These multi-dimensional feature indicators are used to reflect the working status and performance change characteristics of each component of the electrical system. The credibility of the multidimensional feature indicators corresponding to each working condition is evaluated, and their credible evidence is determined. Then, the credibility difference of the multidimensional feature indicators under different working conditions is determined by all the credible evidence. Based on the distribution correlation of multidimensional feature indicators in each operational data, a difference analysis was conducted on the data collected under different working conditions to obtain the consistency differences of multidimensional feature indicators under different working conditions. Based on the credibility difference and the consistency difference, all collected operational data are fused to generate fused feature parameters that can reflect the operating status of the target vehicle's electrical system. The fused feature parameters are then input into the fault diagnosis model to achieve intelligent diagnosis of the fault status of the vehicle's electrical system.

2. The method as described in claim 1, characterized in that, Based on the aforementioned operational data, the extraction of multidimensional feature indicators characterizing the operating status of the electrical system specifically includes: Preprocess the operating data of the target vehicle's electrical system, including voltage, current, power, and output commands from the electronic control unit; Characteristic factors are extracted from the preprocessed operating data to characterize electrical load characteristics, power transmission stability, and electronic control response rate. The various characteristic factors are combined to form a multidimensional characteristic index, which is used to characterize the overall operating status of the target vehicle's electrical system.

3. The method as described in claim 1, characterized in that, The credibility of the multidimensional characteristic indicators corresponding to each working condition is assessed, and the specific evidence for their credibility is determined to include: For multi-dimensional characteristic indicators under various working conditions, a characteristic confidence information set reflecting the fluctuation range, characteristic stability and measurement noise level of the indicators is constructed to quantify the confidence level of each characteristic indicator. Based on the aforementioned feature confidence information set, a multi-source feature credibility assessment model is established according to the DS evidence theory, which transforms confidence information from different feature indicator sources into corresponding evidence support. The evidence support of each feature indicator across different data sources is fused to obtain credible evidence for the corresponding feature indicator under the current working conditions.

4. The method as described in claim 1, characterized in that, The credibility differences of multidimensional feature indicators under different working conditions are determined by all credible evidence, specifically including: A standard credibility matrix is ​​constructed based on all credible evidence to measure the differences in credible distributions between operating conditions. Using the standard credibility matrix, the credibility distribution of each feature index under different working conditions is cross-compared to calculate the evidence support bias and evidence concentration bias between working conditions. Based on the evidence support bias and the evidence set bias, a credibility difference mapping table is constructed to characterize the working condition sensitivity of the feature indicators. Then, the credibility difference of the multidimensional feature indicators under different working conditions is determined through the credibility difference mapping table.

5. The method as described in claim 1, characterized in that, Based on the distribution correlation of multidimensional characteristic indicators in various operational data, a difference analysis was conducted on the data collected under different working conditions. The specific differences in the consistency of multidimensional characteristic indicators under different working conditions include: Extract the distribution correlation between multidimensional feature indicators under various working conditions; Comparative analysis of all distribution correlation degrees was performed to extract the difference information of characteristic correlation structure between various working conditions; Based on the aforementioned difference information, the consistency differences of multidimensional feature indicators under different working conditions are determined.

6. The method as described in claim 1, characterized in that, Based on the aforementioned reliability differences and consistency differences, all collected operational data are fused to generate fused feature parameters that reflect the operational status of the target vehicle's electrical system. Specifically, these parameters include: Construct a fusion weight matrix for the operating data under different working conditions based on the credibility difference and the consistency difference; The fusion weight matrix is ​​used to aggregate multi-dimensional feature indicators under different working conditions to generate a comprehensive feature vector. The comprehensive feature vector is used as input to generate fused feature parameters that can reflect the operating status of the target vehicle's electrical system.

7. The method as described in claim 1, characterized in that, The operating data of the electrical system of the target vehicle under different working conditions are collected by the data acquisition unit on the vehicle.

8. A data analysis system for fault diagnosis of vehicle electrical systems, characterized in that, include: The data acquisition module is used to collect operating data of the electrical system of the target vehicle under different working conditions through intelligent sensors; The feature processing module is used to extract multi-dimensional feature indicators that characterize the operating status of the electrical system based on the operating data. The multi-dimensional feature indicators are used to reflect the working status and performance change characteristics of each component of the electrical system. The feature processing module is also used to evaluate the credibility of the multi-dimensional feature indicators corresponding to each working condition and determine their credible evidence, and then determine the credibility difference of the multi-dimensional feature indicators under different working conditions based on all the credible evidence. The feature processing module is also used to perform difference analysis on the data collected under different working conditions based on the distribution correlation of multidimensional feature indicators in each operating data, and to obtain the consistency difference of multidimensional feature indicators under different working conditions. The feature fusion module is used to fuse all collected operational data based on the confidence difference and the consistency difference to generate fused feature parameters that can reflect the operating status of the target vehicle's electrical system. The fused feature parameters are then input into the fault diagnosis model to achieve intelligent diagnosis of the fault status of the vehicle's electrical system.

9. A computer device comprising a memory and a processor, the memory storing code, characterized in that, The processor is configured to acquire the code and execute the data analysis method for fault diagnosis of a vehicle electrical system as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the data analysis method for fault diagnosis of vehicle electrical systems as described in any one of claims 1 to 7.

Images