Vehicle fault early warning method, device, equipment and platform

Abstract

The invention relates to the technical field of vehicle fault diagnosis, and discloses a vehicle fault early warning method, and the method comprises the steps: obtaining historical fault data, real-time operation data, use scene data and part basic information for a target part; performing statistical feature extraction on the real-time operation data and the use scene data to obtain state statistical features; performing associated feature extraction among the historical fault data, the real-time operation data, the use scene data and the part basic information to obtain associated evolution features; and performing multivariate fusion fault prediction on the target part according to the state statistical characteristics and the associated evolution characteristics to obtain a fusion prediction result, and performing vehicle fault early warning when a failure risk occurs. The method has the beneficial effects that the state statistical characteristics and the associated evolution characteristics are extracted according to the multi-source data to jointly carry out fault prediction and early warning, so that the specificity of the target part is introduced in fault prediction, the accuracy of fault prediction is improved, and the reliability and the safety of the vehicle are further improved.

Description

Technical Field

[0001] This application relates to the field of vehicle fault diagnosis technology, and in particular to a vehicle fault early warning method, device, equipment and platform. Background Technology

[0002] Vehicle components are subjected to continuous dynamic loads and complex stresses during operation, posing a risk of failure and impacting vehicle safety and user experience. Traditional technologies typically address this issue by real-time monitoring of vehicle components and using the monitoring data to predict fault risks, providing proactive warnings and improving vehicle safety. However, these solutions often rely on single-dimensional feature data for prediction. Single-dimensional feature data cannot effectively represent the unique characteristics of each vehicle component and the evolution of faults, thus reducing the accuracy and precision of fault risk prediction and limiting the effectiveness of vehicle maintenance. Therefore, the accuracy of vehicle fault prediction in these technologies still needs improvement. Summary of the Invention

[0003] This application provides a vehicle fault early warning method, device, equipment, and platform. It extracts state statistical features and correlation evolution features from multi-source data of target parts to jointly perform fault prediction and early warning. This introduces the specificity of the target parts into the fault prediction, improves the accuracy of fault prediction, and thus improves the reliability and safety of the vehicle.

[0004] To achieve the above objectives, the main technical solutions adopted in this application include: In a first aspect, embodiments of this application provide a vehicle fault early warning method, the method comprising: For a target part in a vehicle, acquire the target part's historical fault data, real-time operating data, usage scenario data, and basic part information; Statistical features are extracted from the real-time operating data and the usage scenario data to obtain state statistical features related to the target part; Correlation features are extracted between the historical fault data, the real-time operation data, the usage scenario data, and the basic information of the parts to obtain the correlation evolution features of the target part; Based on the state statistical features and the correlation evolution features, multi-dimensional fusion fault prediction is performed on the target part to obtain the fusion prediction result of the target part, and a vehicle fault warning is issued when the fusion prediction result indicates that the target part has a failure risk.

[0005] The vehicle fault early warning method proposed in this application performs real-time monitoring of any target component in a vehicle, acquiring historical fault data, real-time operating data, usage scenario data, and basic component information. Based on the real-time operating data and usage scenario data, it extracts the state statistical features of the target component and extracts the correlation evolution features between different data sets. Based on the state statistical features and correlation evolution features, it predicts the fault condition of the target component and provides a vehicle fault early warning when the target component is at risk of failure. Compared with related technologies, this application extracts features based on multi-source data of the target component, obtaining not only state statistical features characterizing the load behavior of the target component but also correlation evolution features characterizing the wear pattern of the target component. This yields core features that highly contribute to the failure of the target component, improving the quality of component features. Furthermore, by combining state statistical features and correlation evolution features for multi-source fusion fault prediction of the target component, the accuracy of fault prediction is significantly improved, thereby enhancing the reliability and safety of the vehicle.

[0006] Optionally, the step of extracting statistical features from the real-time operating data and the usage scenario data to obtain state statistical features related to the target part includes: Based on the real-time operating data, the working status of the target part is statistically analyzed to obtain the status duration data of the target part; Based on the real-time operating data and the usage scenario data, the target part is statistically analyzed for scenario characteristics to obtain the scenario statistics of the target part; The state statistical features are obtained based on the state duration data and the scene statistical data.

[0007] Optionally, the step of extracting correlation features among the historical fault data, the real-time operating data, the usage scenario data, and the basic information of the parts to obtain the correlation evolution features of the target part includes: Based on the real-time operating data and the usage scenario data, scenario operation analysis is performed on the target part to obtain the scenario loss coefficient of the target part; The performance degradation rate of the target component is obtained by comparing the historical fault data and the real-time operating data. Based on the basic information of the parts, similarity correlation analysis is performed on the real-time operating data to obtain the failure correlation characteristics of the target parts; The correlation evolution characteristics are obtained based on the scenario loss coefficient, the performance degradation rate, and the failure correlation characteristics.

[0008] Optionally, the step of performing scenario operation analysis on the target part based on the real-time operating data and the usage scenario data to obtain the scenario loss coefficient of the target part includes: The real-time operating data and the usage scenario data are grouped by scenario to obtain the usage scenario group of the target part; wherein, the usage scenario group corresponds to a scenario loss weight; Under the aforementioned usage scenario group, the target part is subjected to part loss calculation based on the real-time operating data and the scenario loss weight to obtain the scenario loss coefficient.

[0009] Optionally, the step of performing similarity correlation analysis on the real-time operating data based on the basic information of the part to obtain the failure correlation characteristics of the target part includes: Based on the basic information of the part, perform part similarity analysis on the target part to obtain similar parts to the target part; Based on the historical failure data of the similar parts, the real-time operating data of the target part is correlated and compared to obtain the failure correlation characteristics.

[0010] Optionally, the step of performing multi-factor fusion fault prediction on the target part based on the state statistical features and the correlation evolution features to obtain the fusion prediction result of the target part includes: Based on the state statistical characteristics and the correlation evolution characteristics, a multivariate fault prediction algorithm is used to predict the faults of the target parts respectively, and multivariate fault prediction data of the target parts are obtained. The multivariate fault prediction data is fused to obtain the fused prediction result.

[0011] Optionally, the method further includes: Based on the fusion prediction results, fault similarity matching is performed in the historical case database to obtain historical similar cases related to the fusion prediction results; The historical similar cases are recommended as reference cases for the target part; After the target part is maintained, the maintenance scheme based on the fusion prediction result is coded as a case to obtain a target maintenance case, and the target maintenance case is stored in the historical case library.

[0012] Secondly, embodiments of this application provide a vehicle fault warning device, the device comprising: The parts data acquisition module is used to acquire historical fault data, real-time operating data, usage scenario data and basic information of a target part in a vehicle. The statistical feature extraction module is used to extract statistical features from the real-time operating data and the usage scenario data to obtain state statistical features related to the target part. The correlation feature extraction module is used to extract correlation features between the historical fault data, the real-time operation data, the usage scenario data and the basic information of the parts, so as to obtain the correlation evolution features of the target part; The multi-dimensional fusion prediction module is used to perform multi-dimensional fusion fault prediction on the target part based on the state statistical features and the correlation evolution features, to obtain the fusion prediction result of the target part, and to issue a vehicle fault warning when the fusion prediction result indicates that the target part has a failure risk.

[0013] Thirdly, embodiments of this application provide a computer device, including: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the method described in any of the above embodiments.

[0014] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer instructions, which are used to cause a computer to perform the method described in any one of the above embodiments.

[0015] Fifthly, embodiments of this application provide a computer program product, including computer instructions, which are used to cause a computer to perform the method described in any of the above embodiments. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 This is a step diagram of the vehicle fault warning method provided in the embodiments of this application; Figure 2 This is a logic diagram of multi-level work order distribution in the embodiments of this application; Figure 3 This is a flowchart illustrating the steps of statistical feature extraction in an embodiment of this application; Figure 4 This is a flowchart illustrating the steps of feature extraction in an embodiment of this application. Figure 5 This is a flowchart illustrating the steps of scenario operation analysis in the embodiments of this application; Figure 6 This is a flowchart illustrating the steps of similarity association analysis in the embodiments of this application; Figure 7 This is a flowchart illustrating the steps of multi-element fusion fault prediction in the embodiments of this application; Figure 8 This is a step diagram illustrating the recommended historical fault cases in the embodiments of this application; Figure 9 This is a diagram illustrating the steps involved in constructing the repair case library in this application embodiment; Figure 10 A block diagram of the vehicle fault warning device provided in the embodiments of this application; Figure 11 This is an architecture diagram of the vehicle fault warning system provided in the embodiments of this application; Figure 12 This is a flowchart illustrating the workflow of the dual-mode early warning engine in this application embodiment; Figure 13 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] Vehicle components are subjected to continuous dynamic loads and complex stresses during operation, posing a risk of failure and impacting vehicle safety and user experience. Traditional technologies typically address this issue by real-time monitoring of vehicle components and using the monitoring data to predict fault risks, providing proactive warnings and improving vehicle safety. However, these solutions often rely on single-dimensional feature data for prediction. Single-dimensional feature data cannot effectively represent the unique characteristics of each vehicle component and the evolution of faults, thus reducing the accuracy and precision of fault risk prediction and limiting the effectiveness of vehicle maintenance. Therefore, the accuracy of vehicle fault prediction in these technologies still needs improvement.

[0020] To address the aforementioned issues, this application provides a vehicle fault early warning method, apparatus, device, and platform. For a target component in a vehicle, it acquires historical fault data, real-time operational data, usage scenario data, and basic component information. Statistical features are extracted from the real-time operational data and usage scenario data to obtain state statistical features related to the target component. Correlation features are extracted between the historical fault data, real-time operational data, usage scenario data, and basic component information to obtain correlation evolution features of the target component. Based on the state statistical features and correlation evolution features, multi-dimensional fusion fault prediction is performed on the target component to obtain a fusion prediction result. A vehicle fault early warning is issued when the fusion prediction result indicates a failure risk in the target component.

[0021] The vehicle fault early warning method provided in this application performs real-time monitoring of any target part in a vehicle, acquires historical fault data, real-time operating data, usage scenario data, and basic information of the target part; extracts the state statistical features of the target part based on the real-time operating data and usage scenario data, and extracts the correlation evolution features of the target part between different data; predicts the fault status of the target part based on the state statistical features and correlation evolution features, and provides vehicle fault early warning when the target part is at risk of failure.

[0022] Compared with related technologies, this application extracts features based on multi-source data of the target part, obtaining not only state statistical features characterizing the load behavior of the target part, but also correlation evolution features characterizing the wear pattern of the target part. This yields core features that highly contribute to the failure of the target part, improving the quality of the part features. Based on this, the state statistical features and correlation evolution features are combined to perform multi-source fusion fault prediction of the target part, significantly improving the accuracy of fault prediction and thus enhancing the reliability and safety of the vehicle.

[0023] According to an embodiment of this application, a vehicle fault warning method is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0024] Reference Figure 1 As shown, this embodiment provides a vehicle fault warning method, which includes: S100. For the target part in the vehicle, acquire the target part's historical fault data, real-time operating data, usage scenario data, and basic part information.

[0025] S200. Statistical feature extraction is performed on real-time operation data and usage scenario data to obtain state statistical features related to the target part.

[0026] S300. By extracting correlation features among historical fault data, real-time operation data, usage scenario data, and basic part information, the correlation evolution features of the target part are obtained.

[0027] S400. Based on the state statistical characteristics and correlation evolution characteristics, perform multi-element fusion fault prediction on the target part, obtain the fusion prediction result of the target part, and issue a vehicle fault warning when the fusion prediction result indicates that the target part has a failure risk.

[0028] The historical fault data for the target part can be event records of the target part failing at a specific point in time, or event records of similar parts failing at the same point in time. The historical fault data can be obtained based on the vehicle's entire lifecycle data, including but not limited to part fault type, fault occurrence time, and fault repair records.

[0029] Real-time operating data of the target component can be vehicle operating data associated with the target component during the current driving process, used to indicate the current working status of the target component. Real-time operating data can be acquired through real-time data collection by sensors installed in the vehicle, including but not limited to engine speed, oil temperature, and battery voltage.

[0030] The usage scenario data for the target component can be the vehicle's driving condition data during the current driving process, or the external environmental data in which the vehicle is located, used to represent the operating mode of the target component. Usage scenario data can be acquired through sensors installed inside or outside the vehicle, including but not limited to high-speed driving duration, the proportion of congested road sections, and the frequency of extreme weather. It is understood that both real-time operating data and usage scenario data can be shared remotely via vehicle-to-everything (V2X) communication, allowing vehicle maintenance center staff to monitor them in real time and store them periodically at preset intervals for multi-dimensional fault prediction of the target component.

[0031] The basic information of the target part can be the static design data of the target part, which represents the inherent properties of the target part and is used to provide the individual characteristics that make the target part different from other parts, including but not limited to production batch, material parameters and design service life.

[0032] Specifically, for any target part in a vehicle, real-time data monitoring is performed to acquire historical fault data, real-time operational data, usage scenario data, and basic part information. After acquiring multi-source data for the target part, data preprocessing is performed on each source to improve data quality. Data preprocessing may include steps such as missing value handling, outlier handling, standardization transformation, and data alignment. Missing value handling includes identifying missing values ​​in each source and processing them accordingly, including but not limited to data deletion, interpolation, and statistical imputation. Outlier handling includes identifying outliers in the multi-source data after missing value handling and processing them accordingly, including but not limited to deleting, correcting, or retaining outliers. The standardization process may include: performing format standardization and data structure standardization on multi-source data that has undergone missing value and outlier handling, and performing feature scale normalization on the multi-source data. Format standardization may include time format standardization and numerical representation standardization, while feature scale normalization may include data distribution standardization, data range standardization, and feature encoding. The data alignment process may include: clock skew correction of multi-source data to align the data on the time axis; aligning the sampling frequencies of multi-source data to align the sampling time points, including downsampling of high-frequency data or interpolation sampling of low-frequency data.

[0033] Furthermore, after obtaining multi-source data, statistical features are extracted from the real-time operational data and usage scenario data at preset time periods to obtain the state statistical features related to the target part in the vehicle within any given time period. It should be noted that these state statistical features can represent the basic state and behavioral patterns of the target part during operation, transforming multi-source data into indicators representing the fault state of the target part. This serves as the data foundation for fault prediction of the target part, providing interpretable predictive evidence. Understandably, state statistical features can also be used for preliminary fault diagnosis of the target part. By comparing the state statistical features with a preset threshold range, when the state statistical features deviate from the preset threshold range, a fault can be determined in the target part, thereby enabling rapid diagnosis of the target part's fault condition and improving the efficiency of vehicle fault early warning.

[0034] Furthermore, the multi-source data is segmented according to a preset time period to obtain multi-source data corresponding one-to-one with the time period. Within any given time period, correlation features are extracted between any two multi-source data points to obtain the correlation evolution features of the target part. It should be noted that the correlation evolution features can represent the dynamic correlation trend between multi-source data, which can be used to identify the failure modes and degradation patterns of the target part. This enables periodic monitoring of the target part's health status, early warning when early signs of failure appear, preventing further deterioration of the failure and reducing the failure frequency of the target part.

[0035] Furthermore, in the process of acquiring multi-source data through sensors, to avoid sensor failure leading to data acquisition failure, redundant designs are typically employed for sensors. This results in data redundancy in the multi-source data, meaning that data from multiple sensors may represent the same characteristics. Additionally, during statistical feature extraction and correlation feature extraction, the extracted state statistical features and correlation evolution features can correspond to various types. Some features may represent the same fault condition of the target part, while others may be unrelated to the fault condition of the target part. Considering the above issues, this embodiment, after obtaining the state statistical features and correlation evolution features, also performs feature filtering to remove redundant features and retain features that highly contribute to the fault prediction of the target part, thereby improving the quality of the prediction data and ultimately enhancing the accuracy of fault prediction.

[0036] In some embodiments, the feature selection process may include feature quality selection, feature anomaly selection, and feature type selection. Feature quality selection may be based on the feature quality of state statistical features and correlation evolution features, selecting reliable feature data. It is understood that the reliability of feature data can be determined based on the integrity of multi-source data and the correctness of the feature extraction process. The correctness of the feature extraction process includes whether computational anomalies occur during feature extraction and the timeliness of the feature extraction process. Feature anomaly selection may involve comparing state statistical features and correlation evolution features against a pre-defined normal threshold range. If any feature data deviates from the normal threshold range, that feature data is considered abnormal, potentially indicating a corresponding fault in the target part. Feature type selection may involve pre-determining a set of fault-related features, including feature types related to the fault condition of the target part. State statistical features and correlation evolution features are then compared and selected against this set, identifying feature data unrelated to the fault condition and retaining feature data with high contribution.

[0037] Furthermore, based on the state statistical characteristics and correlation evolution characteristics, multi-dimensional fusion fault prediction is performed on the target part to obtain the fusion prediction result of the target part. It should be noted that the multi-dimensional fusion fault prediction process can involve combining multiple fault prediction algorithms to perform fault prediction separately, and then fusing the respective prediction results to obtain the fusion prediction result. In some embodiments, the multi-dimensional fusion fault prediction process can be implemented through a multi-algorithm fusion model. This model can be a prediction model containing multiple fault prediction algorithms, each operating independently within the model. Each algorithm uses the input state statistical characteristics and correlation evolution characteristics as the prediction basis to perform fault prediction, obtaining its own prediction result. The prediction results of each algorithm are then fused to obtain the fusion prediction result of the target part.

[0038] For example, the multi-algorithm fusion model can be built on the Spark platform. Its fault prediction algorithms can include traditional machine learning algorithms such as Random Forest and Gradient Boosting Decision Tree (GBDT), as well as deep learning models such as LSTM (Long Short-Term Memory) neural networks. Traditional machine learning algorithms can be used to build the basic prediction model and learn complex nonlinear relationships from the input feature data, thus providing highly interpretable prediction results. Meanwhile, deep learning models can be used to capture long-term dependencies in time-series data, extracting temporal dynamic features from the input feature data to perform time evolution analysis on the fault modes of the target part, obtaining time-sensitive prediction results. The prediction results from both can complement each other. By fusing multiple prediction results, a comprehensive characterization of the target part's fault conditions across different dimensions can be achieved, thereby capturing different fault modes of the target part and improving the fault prediction accuracy.

[0039] In some embodiments, the multi-algorithm fusion model can be trained using a sample dataset, which can be obtained by extracting features from historical fault data of the target part and then processing and eliminating feature redundancy. For example, feature redundancy elimination during training can include analysis of variance (ANOVA) and mutual information. The ANOVA process can include: grouping the sample dataset into several sample groups; calculating the variance within each sample group to obtain the within-group variance; calculating the variance between each sample group to obtain the between-group variance; constructing an F-statistic based on the within-group and between-group variances to obtain the F-value and its corresponding F-distribution; calculating the sampling probability based on the obtained F-distribution to obtain the significance probability corresponding to the current F-value; performing significance level analysis on the sample dataset based on the significance probability; and filtering the sample dataset based on the analysis results. The process of using the mutual information method may include: discretizing the sample data in the sample dataset to obtain discrete data; calculating the information entropy of the discrete data to obtain the marginal entropy and joint entropy of the discrete data; calculating the mutual information value of the discrete data based on the marginal entropy and joint entropy; and sorting the sample data corresponding to the discrete data according to the mutual information value to perform feature selection.

[0040] It should be noted that there is a one-to-one correspondence between the multi-algorithm fusion model and the target parts. Different target parts correspond to different multi-algorithm fusion models, and each multi-algorithm fusion model is trained based on the historical fault data of its corresponding target part. The training process of the multi-algorithm fusion model may include: obtaining an initial model; inputting a sample dataset that has undergone feature redundancy removal into the initial model; iteratively training the initial model using the sample dataset; and iteratively optimizing the hyperparameters of the initial model, including decision tree depth, number of neural network iterations, and learning rate, through cross-validation; stopping iteration when the training termination condition is met, and outputting the initial model at this point as the multi-algorithm fusion model. It is understandable that after obtaining the multi-algorithm fusion model, time-series cross-validation can be used to evaluate its performance, determining its prediction accuracy, false alarm rate, and early warning duration, ensuring the model's versatility and stability, and allowing for re-acquisition of the dataset and retraining if the model no longer meets performance requirements. In some embodiments, the multi-algorithm fusion model can also adaptively optimize based on the real-time operating data of the target part and updated maintenance records, thereby continuously improving the accuracy of prediction results by dynamically updating the model parameters of the multi-algorithm fusion model.

[0041] Furthermore, the fusion prediction results can include the failure probability, remaining service life, and potential failure risk level of the target part. The potential failure risk level can be categorized as high-risk, medium-risk, and low-risk. When the failure probability of the target part exceeds a preset failure threshold, or the remaining service life is lower than a preset service life threshold, or the potential failure risk level is high-risk, it indicates that the target part is at risk of failure. A fault warning is then sent to the vehicle, and the warning information, fault cause, and maintenance suggestions are simultaneously pushed to the vehicle maintenance center. This allows staff to quickly and efficiently address the failure risk of the target part, improving maintenance efficiency and reducing the probability of vehicle failure.

[0042] Reference Figure 2 As shown, after obtaining the fusion prediction results, vehicle maintenance work orders are generated based on these results. These work orders are then categorized into three levels based on the potential failure risk level identified in the fusion prediction results: Level 1, Level 2, and Level 3. A Level 1 work order indicates that the target part in the vehicle has a low failure risk and can be handled by the vehicle maintenance center. A Level 2 work order indicates that the target part in the vehicle has a certain failure risk and requires emergency handling. In this case, relevant departments of the vehicle manufacturer, such as service, quality, or R&D, need to intervene to properly address the failure risk of the target part. A Level 3 work order indicates that the target part has a significant failure risk, usually related to vehicle design issues, and requires intervention from the responsible department heads and engineers of the vehicle manufacturer.

[0043] The vehicle fault early warning method provided in this embodiment performs real-time monitoring of any target part in the vehicle, obtains historical fault data, real-time operation data, usage scenario data and basic information of the target part; extracts the state statistical features of the target part based on the real-time operation data and usage scenario data, and extracts the correlation evolution features of the target part between different data; predicts the fault status of the target part based on the state statistical features and correlation evolution features, and provides vehicle fault early warning when the target part is at risk of failure.

[0044] Compared with related technologies, this application extracts features based on multi-source data of the target part, obtaining not only state statistical features characterizing the load behavior of the target part, but also correlation evolution features characterizing the wear pattern of the target part. This yields core features that highly contribute to the failure of the target part, improving the quality of the part features. Based on this, the state statistical features and correlation evolution features are combined to perform multi-source fusion fault prediction of the target part, significantly improving the accuracy of fault prediction and thus enhancing the reliability and safety of the vehicle.

[0045] Reference Figure 3As shown in one embodiment of this application, statistical features are extracted from real-time operating data and usage scenario data to obtain state statistical features related to the target part, including: S210. Based on real-time operating data, perform statistical analysis on the working status of the target part to obtain the status duration data of the target part.

[0046] S220. Based on real-time operation data and usage scenario data, perform scenario feature statistics on the target parts to obtain scenario statistics data for the target parts.

[0047] S230. Obtain state statistical characteristics based on state duration data and scene statistical data.

[0048] Specifically, within the current time period, the working status of the target part is statistically analyzed based on real-time operating data to determine the cumulative time the target part is in a specific working state, thus obtaining the state duration data of the target part, which is used to represent the health evolution and damage consumption of the target part. For example, the state duration data may include features such as the cumulative working time and cumulative start-stop count of the target part within the current time period, as well as the average working time of the target part over multiple time periods.

[0049] Furthermore, within the current time period, based on real-time operational data and usage scenario data, scenario characteristic statistics are performed on the target component to determine the external conditions and working state of the target component, obtaining state duration data of the target component, which is used to represent the workload of the target component under the external conditions. For example, scenario statistical data may include the duration of high engine temperature and engine operating temperature of the vehicle, as well as characteristics such as the operating frequency and workload of the target component under different usage scenarios.

[0050] In some embodiments, the methods for statistical analysis of work status and scene characteristics may include central tendency statistics, dispersion statistics, distribution pattern statistics, and threshold correlation statistics. Central tendency statistics can measure the central tendency of any multi-source data over any given time period, calculating features such as the sum, mean, or median of the data over that time period. Dispersion statistics can measure the variability of any multi-source data over any given time period, calculating features such as the variance or standard deviation. Distribution pattern statistics can measure the shape of the distribution of any multi-source data over any given time period, calculating features such as skewness and kurtosis. Threshold correlation statistics can measure the cumulative anomalies of any multi-source data over any given time period, calculating features such as the duration and frequency of deviations from a preset threshold range.

[0051] Reference Figure 4As shown, in one embodiment of this application, correlation feature extraction is performed between historical fault data, real-time operation data, usage scenario data, and basic part information to obtain the correlation evolution features of the target part, including: S310. Based on real-time operating data and usage scenario data, perform scenario operation analysis on the target part to obtain the scenario loss coefficient of the target part.

[0052] S320. Based on historical fault data and real-time operating data, a performance degradation comparison is performed to obtain the performance degradation rate of the target part.

[0053] S330. Based on the basic information of the parts, perform similarity correlation analysis on the real-time operating data to obtain the failure correlation characteristics of the target parts.

[0054] S340. Obtain the correlation evolution characteristics based on the scenario loss coefficient, performance degradation rate, and failure correlation characteristics.

[0055] Specifically, within the current time period, scenario operation analysis is performed on the target component based on real-time operating data and usage scenario data. A logical relationship is constructed between external environmental conditions and the target component's operating data to obtain the scenario loss coefficient of the target component, which describes the degree to which the target component is affected by the usage scenario. For example, the scenario loss coefficient may include the engine loss coefficient under high-speed driving scenarios and the battery loss coefficient under low-temperature scenarios, where the low-temperature scenario can be a scenario where the vehicle's external ambient temperature is less than -10°C.

[0056] Furthermore, within the current time period, performance degradation is compared based on historical fault data and real-time operating data under similar usage scenarios to determine the performance degradation of the target component at corresponding times, thus obtaining the performance degradation rate of the target component, which represents the rate of deterioration of the target component's health. For example, the performance degradation rate may include the rate of decrease in thermal efficiency of a vehicle engine and the rate of decrease in charge / discharge efficiency of a battery. In some embodiments, a scenario benchmark model can also be constructed for the target component. The scenario benchmark model is trained based on sample data of the target component, enabling it to learn the performance benchmark of the target component under different usage scenarios. After obtaining the scenario benchmark model, based on the usage scenario corresponding to the real-time operating data, the operating performance benchmark of the target component under that usage scenario is output through the scenario benchmark model, and the performance degradation is compared between the real-time operating data and the operating performance benchmark to obtain the performance degradation rate of the target component.

[0057] Furthermore, within the current time period, similarity correlation analysis is performed on real-time operational data based on the basic information of the parts. This allows for analogical analysis of the target part based on data from similar parts, yielding failure correlation characteristics of the target part. These characteristics represent the common failure features of the part type to which the target part belongs. For example, failure correlation characteristics may include the health index and co-occurrence frequency of the part type to which the target part belongs.

[0058] Reference Figure 5 As shown, in one embodiment of this application, scenario operation analysis is performed on the target part based on real-time operation data and usage scenario data to obtain the scenario loss coefficient of the target part, including: S312. Group the real-time operation data and usage scenario data into scenarios to obtain the usage scenario group of the target part; among them, the usage scenario group has a corresponding scenario loss weight.

[0059] S314. Under the usage scenario group, calculate the component loss of the target component based on real-time running data and scenario loss weight to obtain the scenario loss coefficient.

[0060] Specifically, based on the scenario conditions corresponding to the usage scenario data, the usage scenario data is grouped into scenarios, resulting in scenario data corresponding to each scenario condition. Based on the scenario conditions, the operational data corresponding to each scenario condition is extracted from the real-time operational data, and this operational data is associated with the corresponding usage scenario group to obtain the usage scenario group for the target part. Based on the business logic relationship between the operational data and scenario conditions in the usage scenario group, the scenario loss weight of the target part under the corresponding scenario conditions is defined. In essence, the usage scenario group represents the total workload of the target part under the corresponding scenario conditions.

[0061] Furthermore, for any usage scenario group, the component loss is calculated based on the real-time operating data of the target component in that usage scenario group and the scenario loss weight, to obtain the scenario loss coefficient of the target component under the corresponding scenario conditions. For example, in a high-speed driving scenario, the loss weight of the engine can be 0.8; in a low-temperature scenario, the loss weight of the battery can be 1.2.

[0062] Understandably, after obtaining the scenario loss coefficient, a correlation analysis can be performed between the scenario loss coefficient and the actual failure rate of the target part. If the scenario loss coefficient is high but the correlation with the actual failure rate is low, the loss weight can be optimized or the parameter combination of the operating data can be adjusted to improve the accuracy of the scenario loss coefficient.

[0063] Reference Figure 6As shown, in one embodiment of this application, based on the basic information of the part, a similarity correlation analysis is performed on the real-time operating data to obtain the failure correlation characteristics of the target part, including: S332. Based on the basic information of the parts, perform part similarity analysis on the target parts to obtain similar parts to the target parts.

[0064] S334. Based on the historical failure data of similar parts, compare and correlate the real-time operating data of the target part to obtain failure correlation characteristics.

[0065] Specifically, design data and individual characteristics of the target part are obtained based on the part's basic information to perform part similarity analysis, thereby identifying similar parts and forming a similarity comparison set. It can be understood that all parts in the similarity comparison set correspond to similar part basic information; these could be identical parts from the same production line or identical parts used in the same location. Similar parts also have historical fault data, which can be obtained through data collection based on the working process of similar parts, or through part testing and group statistical analysis of a specific number of similar parts.

[0066] Furthermore, after obtaining similar parts and their corresponding historical fault data, data alignment and standardization preprocessing are performed on the historical fault data of similar parts and the real-time operating data of the target part to ensure that both are based on the same data foundation. After preprocessing, correlation analysis is performed on the historical fault data of similar parts and the real-time operating data of the target part to obtain the data deviation between the two.

[0067] It should be noted that data deviation can represent the fault correlation between the target part and similar parts. The smaller the data deviation, the higher the correlation between the historical fault data of similar parts and the real-time operating data of the target part, indicating that the historical fault data of similar parts has an important reference value for the fault prediction of the target part. Conversely, the larger the data deviation, the lower the correlation between the historical fault data of similar parts and the real-time operating data of the target part, indicating that the historical fault data of similar parts has little reference value for the fault prediction of the target part. Therefore, the correlation weights corresponding to the historical fault data of each similar part can be determined based on the data deviation. By weighting and fusing the historical fault data of each similar part according to the correlation weights, the failure correlation characteristics of the target part can be obtained.

[0068] It is understood that in this embodiment, by identifying similar parts to the target part and weighting and fusing the historical fault data of the similar parts according to the correlation between the similar parts and the target part, failure association features are obtained. This transforms the individual prediction for the target part into a group prediction for the corresponding part type, thereby improving the ability of the failure association features to express the failure status of the target part and thus improving the prediction efficiency and accuracy.

[0069] Reference Figure 7 As shown, in one embodiment of this application, based on state statistical characteristics and correlation evolution characteristics, a multi-element fusion fault prediction is performed on the target part to obtain the fusion prediction result of the target part, including: S410. Based on the state statistical characteristics and correlation evolution characteristics, a multivariate fault prediction algorithm is used to predict the faults of the target parts respectively, and multivariate fault prediction data of the target parts are obtained.

[0070] S420. Perform multi-factor fusion on the multi-factor fault prediction data to obtain the fusion prediction result.

[0071] Specifically, the multivariate fault prediction algorithm can be a variety of fault prediction algorithms included in the multi-algorithm fusion model, including but not limited to traditional machine learning algorithms such as random forest and gradient boosting tree, as well as deep learning models such as LSTM neural network. Each fault prediction algorithm is independent of each other in the multivariate fault prediction model, and each uses the input state statistical features and correlation evolution features as the prediction basis to perform fault prediction and obtain its own prediction results as multivariate fault prediction data.

[0072] In some embodiments, before fault prediction, feature fusion can be performed on state statistical features and correlation evolution features to obtain multi-source fused features for fault prediction. The feature fusion process may include: normalizing the state statistical features and correlation evolution features respectively to unify their data volume; merging the normalized state statistical features and correlation evolution features to obtain a unified feature matrix, which represents the basic state and deep fault patterns of the target part in all dimensions; removing redundancy from the unified feature matrix, selecting core features that contribute highly to fault prediction, and removing duplicate or strongly correlated features to obtain multi-source fused features.

[0073] Furthermore, the multivariate fault prediction data is fused to achieve complementary advantages among each fault prediction algorithm, resulting in a fused prediction result. In some embodiments, the multivariate fusion process can employ an ensemble learning approach. First, a meta-learner performs trust learning on multiple fault prediction algorithms to obtain the trust weights for each algorithm. Then, the multivariate fault prediction data is weighted and fused based on these trust weights to obtain a fused prediction result, thereby reconstructing the actual fault condition of the target part from the multivariate fault prediction data.

[0074] Understandably, traditional machine learning algorithms can be used to build basic predictive models and learn complex nonlinear relationships from input feature data, thus providing highly interpretable prediction results. Deep learning models, on the other hand, can capture long-term dependencies in time-series data, extracting temporal dynamic features from input feature data to perform time evolution analysis of the failure modes of target parts, obtaining time-sensitive prediction results. The prediction results of both can complement each other; by fusing multiple prediction results, a comprehensive characterization of the failure status of the target part across different dimensions can be achieved, thereby capturing different failure modes of the target part and improving the accuracy of failure prediction.

[0075] Reference Figure 8 As shown, in one embodiment of this application, the method further includes: S510. Based on the fusion prediction results, perform fault similarity matching in the historical case database to obtain historical similar cases related to the fusion prediction results.

[0076] S520. Recommend relevant historical similar cases as reference cases for the target part.

[0077] S530. After the target part is maintained, the maintenance plan based on the fusion prediction results is coded as a case to obtain the target maintenance case, and the target maintenance case is stored in the historical case library.

[0078] Specifically, after obtaining the fusion prediction result of the target part, the fusion prediction result is vector-encoded to obtain the target feature vector representing the fusion prediction result. Similarity is calculated based on the target feature vector and the case feature vectors of each historical case in the historical case library, thereby accurately quantifying the distance between the two in the feature space and obtaining the case similarity between the target feature vector and each case feature vector. Based on the case similarity, the historical cases in the historical case library are ranked, and several historical cases with high similarity are selected as historical similar cases related to the fusion prediction result. For example, the similarity calculation process can use a weighted cosine similarity method, assigning higher weights to features strongly related to the cause of the fault, such as the operating parameters of the core component and fault alarm codes, in both the target feature vector and the case feature vectors of historical cases, and then using the cosine similarity algorithm to calculate the case similarity between the two, thereby effectively improving the accuracy of the similarity calculation and avoiding interference from irrelevant features in the matching results.

[0079] Furthermore, similar historical cases are recommended to the corresponding engineers according to the case recommendation strategy. The engineers then judge the reference value of the similar historical cases and use the ones with high reference value as reference cases for the target parts to help the engineers quickly formulate maintenance plans for the target parts in order to maintain them.

[0080] Furthermore, after maintaining the target part based on the maintenance plan and resolving the fault conditions represented by the fusion prediction results, Natural Language Processing (NLP) technology is used to extract knowledge entities from the maintenance plan, obtaining the plan's knowledge entities. Entity associations are then extracted between these knowledge entities to obtain the plan entity associations. Based on standardized case templates, cases are structured according to the plan knowledge entities and their associations, forming target maintenance cases which are stored in a historical case repository. This ensures the standardization and reusability of information, laying the foundation for knowledge accumulation.

[0081] In some embodiments, knowledge statistics techniques can be used to cluster and analyze historical cases in the historical case library, statistically analyzing case indicators such as application frequency, success rate, implementation cost, and applicable scenario coverage for each historical case. This allows for the selection of the optimal case from all historical cases to resolve the corresponding fault. For the optimal solution among the optimal cases, this optimal solution is standardized, clarifying fault judgment criteria, preparatory work, step-by-step operation procedures, key precautions, effectiveness verification methods, and applicable scenario scope, forming a standardized processing template. After obtaining the standardized processing template, senior engineers verify its rationality and rigor. Upon approval, the standardized processing template is included in the standard template library, assigning a unique identifier and version number to each template for standardized management. Subsequent iterations can be made based on actual application needs to ensure continuous template adaptation.

[0082] Reference Figure 9 As shown, historical cases in the historical case library can be obtained by extracting and structuring knowledge from historical maintenance plans. For any historical maintenance plan, natural language processing technology is used to extract knowledge entities, resulting in the plan's knowledge entities. Entity associations are then extracted between these knowledge entities to obtain the plan entity associations. Based on standardized case templates in the standard template library, cases are structured according to the plan knowledge entities and their associations, forming historical cases which are then stored in the historical case library, thereby maximizing knowledge value. It is understood that the historical case library also has adaptive optimization capabilities, dynamically optimizing the case recommendation strategy based on the practical application effects of each historical case to ensure the practicality and timeliness of the knowledge.

[0083] Accordingly, please refer to Figure 10 This application provides a vehicle fault warning device, which includes: The parts data acquisition module 1010 is used to acquire historical fault data, real-time operating data, usage scenario data and basic information of target parts in a vehicle.

[0084] The statistical feature extraction module 1020 is used to extract statistical features from real-time operating data and usage scenario data to obtain state statistical features related to the target part.

[0085] The correlation feature extraction module 1030 is used to extract correlation features between historical fault data, real-time operation data, usage scenario data and basic part information to obtain the correlation evolution features of the target part.

[0086] The multi-element fusion prediction module 1040 is used to perform multi-element fusion fault prediction on the target part based on state statistical characteristics and correlation evolution characteristics, obtain the fusion prediction result of the target part, and provide vehicle fault warning when the fusion prediction result indicates that the target part has a failure risk.

[0087] In some optional implementations, the statistical feature extraction module 1020 includes: The working status statistics unit is used to perform working status statistics on the target part based on real-time operating data, and obtain the status duration data of the target part.

[0088] The scene feature statistics unit is used to perform scene feature statistics on the target part based on real-time operation data and usage scenario data, and obtain scene statistics data of the target part.

[0089] The statistical feature acquisition unit is used to obtain state statistical features based on state duration data and scene statistical data.

[0090] In some optional implementations, the association feature extraction module 1030 includes: The scenario operation analysis unit is used to perform scenario operation analysis on the target part based on real-time operation data and usage scenario data, and obtain the scenario loss coefficient of the target part.

[0091] The performance degradation comparison unit is used to compare the performance degradation based on historical fault data and real-time operating data to obtain the performance degradation rate of the target part.

[0092] The similarity association analysis unit is used to perform similarity association analysis on real-time operating data based on the basic information of the parts, and to obtain the failure association characteristics of the target parts.

[0093] The correlation feature acquisition unit is used to obtain correlation evolution features based on the scene loss coefficient, performance degradation rate and failure correlation features.

[0094] In some optional implementations, the scenario execution analysis unit includes: The scenario grouping subunit is used to group real-time running data and usage scenario data into scenario groups to obtain the usage scenario groups of the target parts; among them, the usage scenario groups correspond to scenario loss weights.

[0095] The loss calculation subunit is used to calculate the loss of the target part based on real-time operating data and scene loss weight under the usage scenario group, and obtain the scene loss coefficient.

[0096] In some optional implementations, the similarity association analysis unit includes: The part similarity analysis subunit is used to perform part similarity analysis on the target part based on the part's basic information and to obtain similar parts to the target part.

[0097] The correlation comparison subunit is used to compare the real-time operating data of the target part with the historical failure data of similar parts to obtain failure correlation characteristics.

[0098] In some optional implementations, the multivariate fusion prediction module 1040 includes: The multivariate fault prediction unit is used to perform fault prediction on the target parts according to the state statistical characteristics and correlation evolution characteristics, and to obtain multivariate fault prediction data of the target parts.

[0099] The multivariate result fusion unit is used to perform multivariate fusion on multivariate fault prediction data to obtain fused prediction results.

[0100] In some alternative implementations, the device also includes a maintenance case recommendation module, comprising: The fault similarity matching unit is used to perform fault similarity matching in the historical case library based on the fusion prediction results, and obtain historical similar cases related to the fusion prediction results.

[0101] The case-related recommendation unit is used to recommend similar historical cases as reference cases for the target part.

[0102] The case coding storage unit is used to code the maintenance plan based on the fusion prediction results after the target part has been maintained, obtain the target maintenance case, and store the target maintenance case in the historical case library.

[0103] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0104] In this embodiment, the vehicle fault warning device is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0105] This application also provides a vehicle fault early warning platform, comprising a user layer, a functional layer, a technical layer, and a data layer connected in sequence; wherein, the functional layer is used to execute the method described in any one of the above embodiments.

[0106] Reference Figure 11As shown, the vehicle fault early warning platform includes a user layer, a functional layer, a technical layer, and a data layer. The user layer integrates the entire service process of the vehicle product line, regional service departments, and vehicle maintenance centers. When any target part in the vehicle experiences a fault risk, the user layer can distribute the fault information and repair work order to all ports, realizing end-to-end collaboration from the production end to the service end. The functional layer can be used to execute the methods described in any of the above embodiments.

[0107] Reference Figure 12 As shown, the technology layer employs a big data processing technology stack, including a stream processing framework based on Kafka and Flink, a batch processing pipeline built on Spark, and a task scheduling system based on DolphinScheduler. The Kafka and Flink stream processing framework connects to the vehicle network to obtain real-time data streams from each vehicle, generating multi-source data. The Spark-based batch processing pipeline performs offline data analysis. The DolphinScheduler-based task scheduling system enables intelligent scheduling of complex tasks. The data layer connects to multiple data sources, including but not limited to the vehicle network, a historical case library, and a key quality assessment system. The vehicle network provides real-time data streams from each vehicle, allowing for real-time monitoring of their operational status. The historical case library stores historical cases that can serve as reference cases for repairing target parts. The key quality assessment system can be customized based on each company's quality indicators to describe the fault conditions of various target parts in the vehicle.

[0108] Please see Figure 13 , Figure 13 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application, such as... Figure 13 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 13 Take a processor 10 as an example.

[0109] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0110] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0111] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0112] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0113] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0114] This application also provides a computer-readable storage medium. The methods described in this application can be implemented in hardware or firmware, or implemented as recordable on a storage medium, or implemented as computer code downloaded over a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and subsequently stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the methods shown in the above embodiments are implemented.

[0115] This application provides a computer program product including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the method of any embodiment of this application.

[0116] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and such modifications and variations all fall within the scope defined by the appended claims.

[0117] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, a computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0118] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0119] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0120] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1A device that provides the functions specified in one or more boxes.

[0121] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0122] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0123] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0124] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0125] The above description is merely an embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of this application should be included within the scope of the claims of this application.

[0126] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A vehicle fault early warning method, characterized in that, The method includes: For a target part in a vehicle, acquire the target part's historical fault data, real-time operating data, usage scenario data, and basic part information; Statistical features are extracted from the real-time operating data and the usage scenario data to obtain state statistical features related to the target part; Correlation features are extracted between the historical fault data, the real-time operation data, the usage scenario data, and the basic information of the parts to obtain the correlation evolution features of the target part; Based on the state statistical features and the correlation evolution features, multi-dimensional fusion fault prediction is performed on the target part to obtain the fusion prediction result of the target part, and a vehicle fault warning is issued when the fusion prediction result indicates that the target part has a failure risk.

2. The method according to claim 1, characterized in that, The step of extracting statistical features from the real-time operating data and the usage scenario data to obtain state statistical features related to the target part includes: Based on the real-time operating data, the working status of the target part is statistically analyzed to obtain the status duration data of the target part; Based on the real-time operating data and the usage scenario data, the target part is statistically analyzed for scenario characteristics to obtain the scenario statistics of the target part; The state statistical features are obtained based on the state duration data and the scene statistical data.

3. The method according to claim 1, characterized in that, The step of extracting correlation features among the historical fault data, the real-time operation data, the usage scenario data, and the basic information of the parts to obtain the correlation evolution features of the target part includes: Based on the real-time operating data and the usage scenario data, scenario operation analysis is performed on the target part to obtain the scenario loss coefficient of the target part; The performance degradation rate of the target component is obtained by comparing the historical fault data and the real-time operating data. Based on the basic information of the parts, similarity correlation analysis is performed on the real-time operating data to obtain the failure correlation characteristics of the target parts; The correlation evolution characteristics are obtained based on the scenario loss coefficient, the performance degradation rate, and the failure correlation characteristics.

4. The method according to claim 3, characterized in that, The step of performing scenario operation analysis on the target part based on the real-time operation data and the usage scenario data to obtain the scenario loss coefficient of the target part includes: The real-time operating data and the usage scenario data are grouped by scenario to obtain the usage scenario group of the target part; wherein, the usage scenario group corresponds to a scenario loss weight; Under the aforementioned usage scenario group, the target part is subjected to part loss calculation based on the real-time operating data and the scenario loss weight to obtain the scenario loss coefficient.

5. The method according to claim 3, characterized in that, The step of performing similarity correlation analysis on the real-time operating data based on the basic information of the part to obtain the failure correlation characteristics of the target part includes: Based on the basic information of the part, perform part similarity analysis on the target part to obtain similar parts to the target part; Based on the historical failure data of the similar parts, the real-time operating data of the target part is correlated and compared to obtain the failure correlation characteristics.

6. The method according to claim 1, characterized in that, The step of performing multi-factor fusion fault prediction on the target part based on the state statistical features and the correlation evolution features to obtain the fusion prediction result of the target part includes: Based on the state statistical characteristics and the correlation evolution characteristics, a multivariate fault prediction algorithm is used to predict the faults of the target parts respectively, and multivariate fault prediction data of the target parts are obtained. The multivariate fault prediction data is fused to obtain the fused prediction result.

7. The method according to claim 1, characterized in that, The method further includes: Based on the fusion prediction results, fault similarity matching is performed in the historical case database to obtain historical similar cases related to the fusion prediction results; The historical similar cases are recommended as reference cases for the target part; After the target part is maintained, the maintenance scheme based on the fusion prediction result is coded as a case to obtain a target maintenance case, and the target maintenance case is stored in the historical case library.

8. A vehicle fault warning device, characterized in that, The device includes: The parts data acquisition module is used to acquire historical fault data, real-time operating data, usage scenario data and basic information of a target part in a vehicle. The statistical feature extraction module is used to extract statistical features from the real-time operating data and the usage scenario data to obtain state statistical features related to the target part. The correlation feature extraction module is used to extract correlation features between the historical fault data, the real-time operation data, the usage scenario data and the basic information of the parts, so as to obtain the correlation evolution features of the target part; The multi-dimensional fusion prediction module is used to perform multi-dimensional fusion fault prediction on the target part based on the state statistical features and the correlation evolution features, to obtain the fusion prediction result of the target part, and to issue a vehicle fault warning when the fusion prediction result indicates that the target part has a failure risk.

9. A computer device, characterized in that, include: A memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, the processor executing the computer instructions to perform the method of any one of claims 1 to 7.

10. A vehicle fault early warning platform, characterized in that, It includes a user layer, a functional layer, a technical layer, and a data layer connected in sequence; wherein the functional layer is used to perform the method of any one of claims 1 to 7.

Images