Artificial intelligence-based automobile part failure prediction method and system

Abstract

The application provides a kind of based on artificial intelligence's automobile parts fault prediction method and system, it is related to artificial intelligence technical field, first linkage intelligent driving system's environment perception and parts state monitoring module constructs multi-component dynamic coupling modeling data, analyzes the chain transmission logic of fault between core components Generation transmission path data, fusion real-time environment and multi-component operating state data, with transmission path data dynamically associated to generate real-time coupling state data, according to real-time coupling state and transmission path data deduce fault evolution process to generate prediction result, utilize the corresponding relationship of real-time operating state and prediction result Iterative update modeling data, realize the dynamic adaptation of model to intelligent driving condition.The application comprehensively considers multiple factors, accurately predicts component failure, and ensures intelligent driving safety.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and more specifically, to an artificial intelligence-based method and system for predicting automotive component failures. Background Technology

[0002] In the process of automotive intelligence development, intelligent driving systems are becoming increasingly widespread, and fault prediction of automotive components is crucial for ensuring driving safety and reducing maintenance costs. Currently, there are many shortcomings in methods for predicting automotive component faults.

[0003] Some traditional methods focus only on monitoring the independent operating status of individual components, neglecting the complex dynamic coupling relationships between components in intelligent driving scenarios. During intelligent driving, different components influence and interact with each other. For example, the information acquired by the environmental perception module of intelligent driving affects the operation of multiple core components such as the power system and braking system. Traditional methods cannot capture the correlation between these multiple components, leading to inaccurate fault prediction.

[0004] Furthermore, most existing methods do not fully integrate real-time environmental data from intelligent driving. The intelligent driving environment is complex and ever-changing, including different road conditions and weather conditions. These factors significantly affect the operating status of components and the probability of failure. The lack of utilization of real-time environmental data makes fault prediction unable to adapt to the actual operating conditions of intelligent driving, making it difficult to accurately predict component failures in advance, and failing to meet the high safety and reliability requirements of intelligent driving. Summary of the Invention

[0005] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide an artificial intelligence-based method for predicting automotive component failures, the method comprising: The environmental perception module and component status monitoring module of the intelligent driving system are linked to construct dynamic coupling modeling data of multiple components in intelligent driving scenarios. Based on the dynamic coupling modeling data of multiple components in the intelligent driving scenario, the chain transmission logic of faults among the core components of intelligent driving is analyzed, and the chain transmission path data of faults in intelligent driving components is generated. By integrating real-time environmental data of intelligent driving with real-time operating status data of multiple components, and dynamically associating it with fault chain transmission path data of intelligent driving components, real-time coupled status data of multiple intelligent driving components is generated. Based on the real-time coupled status data of multiple intelligent driving components and the fault chain transmission path data of intelligent driving components, the evolution process of faults under intelligent driving conditions is deduced, and the fault prediction results of intelligent driving components are generated. By leveraging the correspondence between real-time operating status data of multiple intelligent driving components and fault prediction results of these components, the dynamic coupling modeling data of multiple components in intelligent driving scenarios is iteratively updated, enabling the artificial intelligence model to dynamically adapt to intelligent driving conditions.

[0006] Furthermore, embodiments of the present invention also provide an artificial intelligence-based automotive component fault prediction system, characterized in that it includes: A processor; a machine-readable storage medium for storing machine-executable instructions of the processor; wherein the processor is configured to execute the aforementioned AI-based automotive component fault prediction method by executing the machine-executable instructions.

[0007] In another aspect, embodiments of the present invention also provide a computer program product, the computer program product including machine-executable instructions, the machine-executable instructions being stored in a computer-readable storage medium, the processor of the AI-based automotive component fault prediction system reading the machine-executable instructions from the computer-readable storage medium, the processor executing the machine-executable instructions, causing the AI-based automotive component fault prediction system to execute the aforementioned AI-based automotive component fault prediction method.

[0008] Based on the above, by linking the environmental perception module and the component status monitoring module of the intelligent driving system, dynamic coupling modeling data of multiple components is constructed. This captures the complex dynamic relationships between various components in the intelligent driving scenario. Then, the chain transmission logic of faults among core components is analyzed and transmission path data is generated, revealing the pattern of fault propagation. This integrates real-time environmental data of intelligent driving with real-time operating status data of multiple components, and dynamically correlates it with fault chain transmission path data to generate real-time coupled status data. This fully considers the complexity and variability of the intelligent driving environment. Based on these real-time coupled status data, the fault evolution process is deduced to generate prediction results, achieving accurate prediction of component faults. The correspondence between real-time operating status data and prediction results is used to iteratively update the dynamic coupling modeling data, enabling the artificial intelligence model to continuously adapt to changes in intelligent driving conditions, continuously improve the accuracy and reliability of fault prediction, and effectively ensure the safety and stability of intelligent driving. Attached Figure Description

[0009] Figure 1 This is a schematic diagram of the execution flow of the artificial intelligence-based automotive component fault prediction method provided in this embodiment of the invention.

[0010] Figure 2 This is a schematic diagram of exemplary hardware and software components of an artificial intelligence-based automotive component fault prediction system provided in an embodiment of the present invention. Detailed Implementation

[0011] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating an artificial intelligence-based method for predicting automotive component failures according to an embodiment of the present invention. The flowchart shows the main steps of steps S110, S120, S130, S140, and S150. The artificial intelligence-based method for predicting automotive component failures will be described in detail below.

[0012] Step S110: Link the environmental perception module and component status monitoring module of the intelligent driving system to construct dynamic coupling modeling data of multiple components in the intelligent driving scenario.

[0013] In this embodiment, taking the driving scenario of an intelligent driving vehicle in urban road autonomous driving mode as an example, the system constructs dynamic coupling modeling data of multiple components in the intelligent driving scenario by linking the environmental perception module and the component status monitoring module of the intelligent driving system. This process requires comprehensive consideration of environmental factors, component operating status, historical fault data, and other information.

[0014] Step S120: The environmental perception module of the intelligent driving system collects environmental data during the intelligent driving process. The environmental data during the intelligent driving process includes road type data, traffic flow data, meteorological condition data, lighting condition data, and obstacle distribution data.

[0015] In the aforementioned urban road autonomous driving scenario, the environmental perception module of the intelligent driving system begins to function. This module uses various sensors installed on the vehicle, such as cameras, radar, and lidar, to collect real-time data on road type (e.g., whether the current road is a main road, secondary road, or secondary road); traffic flow data (e.g., the number of vehicles and pedestrians passing through a specific road section per unit time); meteorological conditions (e.g., real-time temperature, humidity, rainfall, wind direction, and wind speed); lighting conditions (e.g., daytime light intensity, backlighting, and nighttime streetlight brightness); and obstacle distribution data (e.g., the location, size, and movement of other vehicles, pedestrians, non-motorized vehicles, buildings, traffic signs, and markings around the vehicle). This environmental data is stored in a multi-dimensional array, with each dimension corresponding to an environmental parameter, and the elements in the array representing the specific values ​​of that parameter at different sampling times.

[0016] In this embodiment, step S120 may include the following sub-steps: Step S121: The component status monitoring module of the linkage intelligent driving system collects the operating status data of the core components of intelligent driving. The operating status data of the core components of intelligent driving includes steering system component data, braking system component data, power system component data, perception system component data, and decision control system component data. Each system component data includes operating temperature data, operating pressure data, vibration data, energy consumption data, running time data, and control signal interaction data.

[0017] Following this, the component status monitoring module also simultaneously begins data acquisition. For steering system components, it collects data on the steering gear's operating temperature (the temperature of the internal hydraulic fluid or motor); operating pressure data (such as the output pressure of the power steering pump); vibration data (the vibration frequency and amplitude obtained through vibration sensors installed on the steering gear); energy consumption data (the real-time power consumption of the steering system motor); runtime data (the cumulative operating time of the steering system from vehicle startup); and control signal interaction data (signals sent from the steering wheel angle sensor to the ECU and control signals fed back from the ECU to the steering actuator). The data acquisition methods for braking system components, powertrain components, sensing system components, and decision control system components are similar, collecting the aforementioned data from key components within their respective systems. This operational status data is also stored in the form of multi-dimensional arrays, with each system component corresponding to an independent array. Different dimensions of the array correspond to different operational status parameters, and the array elements are the values ​​of the corresponding parameters at each sampling time.

[0018] Step S122: Collect historical fault data corresponding to core components of intelligent driving. The historical fault data corresponding to core components of intelligent driving includes fault type data, intelligent driving mode data when the fault occurs, environmental data when the fault occurs, component link data involved in the fault, and control response change data induced by the fault.

[0019] When constructing modeling data, it is also necessary to collect historical fault data corresponding to core components of intelligent driving. This data is obtained from the vehicle's fault diagnosis system and cloud database. The fault type data clarifies whether the fault is mechanical, electrical, or signal fault, etc.; the intelligent driving mode data at the time of the fault, recording whether the vehicle was in highway autonomous driving mode, urban road autonomous driving mode, etc., at the time of the fault; environmental data at the time of the fault, i.e., the road, traffic, weather, and other environmental conditions at that time; component link data involved in the fault, specifically referring to the list of components that have shown abnormal states and their upstream and downstream connections in the subsystem, recorded on-site by fault codes and on-board diagnostic systems at the time of the fault. This data is not a pre-known complete propagation path, but a snapshot record of the affected components after the fault occurs, used for subsequent mining of fault propagation patterns; and control response change data induced by the fault, reflecting the adjustment of the intelligent driving system's control strategy after the fault occurs, such as reduced power output and limited steering angle. The above historical fault data is stored in the form of a structured data table, with each row representing a fault instance and each column corresponding to a fault-related parameter.

[0020] Step S123: Classify the operating status data of the core components of intelligent driving according to the intelligent driving mode. The intelligent driving modes include highway automatic driving mode, urban road automatic driving mode, traffic jam following mode, curve driving mode, and parking mode, and generate component operating status data classified by mode.

[0021] After acquiring the operational status data of core intelligent driving components, it is necessary to classify them according to the intelligent driving mode. First, current and historical intelligent driving mode information is obtained from the vehicle's control system. Then, based on this mode information, the component operational status data is grouped. For example, all component operational status data collected in highway autonomous driving mode is grouped into one category, data from urban road autonomous driving mode is grouped into another, and so on, processing data in congestion following mode, curve driving mode, and parking mode respectively. After classification, component operational status data categorized by mode is generated, with each mode corresponding to a data subset containing various operational status data of all components under that mode.

[0022] Step S124: Correlate the environmental data during the intelligent driving process with the component operating status data classified by mode, explore the correlation of changes in the operating status of each component under different environmental conditions, and generate environment-component correlation data.

[0023] Next, environmental data during intelligent driving is correlated with component operating status data categorized by mode. Taking urban road autonomous driving mode as an example, component operating status data in this mode is matched with corresponding environmental data, ensuring that each set of component operating status data has a corresponding environmental data. Then, data mining algorithms are used to analyze the changes in the operating status parameters of each component under different road types, traffic flow, weather conditions, and other environmental conditions. For example, the trends in energy consumption and operating temperature data of powertrain components in congested urban road environments are analyzed, along with their correlation with other component operating status parameters. Through the above analysis, environment-component correlation data is generated, presented in the form of correlation rules. Each rule represents the correspondence between a combination of environmental conditions and changes in component operating status.

[0024] Step S125: Extract the sequence of changes in the operating state of components before the occurrence of the fault from the historical fault data corresponding to the core components of intelligent driving, and generate the fault pre-fault state change sequence data.

[0025] From historical fault data, for each fault instance, the operating status data of components in the period preceding the fault is extracted. Taking a steering system fault as an example, the changes in operating status parameters such as operating temperature, operating pressure, and vibration of the steering system and related components in the 30 minutes before the fault occurs are extracted to form a time series. These time series corresponding to all fault instances are integrated to generate a fault pre-fault state change sequence data. This fault pre-fault state change sequence data is stored in the form of a three-dimensional array. The first dimension represents different fault instances, the second dimension represents the sampling points of the time series, and the third dimension represents the operating status parameters of the components.

[0026] Step S126: Integrate the component operation status data, environment-component association data, and fault pre-state change sequence data classified by mode to mine the coupling relationship of multiple component operation status under intelligent driving mode and generate multi-component coupling relationship data.

[0027] In the aforementioned urban road autonomous driving scenario, component operating status data, environment-component correlation data, and fault pre-fault state change sequence data categorized by mode are fused. Specifically, a feature-level fusion algorithm based on a multi-layer attention mechanism is employed, with the following steps: First, data preprocessing is performed, cleaning the three types of time-series data to remove missing and outlier values, followed by Z-score standardization to ensure a mean of 0 and a variance of 1 for each class. Next, three parallel, structurally identical temporal convolutional networks are used to extract features from the three preprocessed data classes, with each network outputting a 1024-dimensional feature vector. Then, these three 1024-dimensional feature vectors are input into a shared fully connected layer to obtain three scalar attention scores, which are then normalized using the Softmax function to obtain three weight coefficients, the sum of which is 1. Finally, the final fused feature vector is calculated as the weighted sum of each weight coefficient multiplied by its corresponding feature vector. This fused feature vector serves as the unified data representation for subsequent mining of multi-component coupling relationships. Next, data mining methods such as association analysis and cluster analysis are used to uncover the mutual influence relationships between the operating states of different components. For example, in urban road autonomous driving mode, the analysis examines whether the operating pressure of braking system components changes when the operating temperature of powertrain components increases, and the patterns and extent of such changes. Through this mining, multi-component coupling relationship data is generated and stored in the form of a graph structure. Nodes in the graph represent components, edges represent the coupling relationships between components, and the weights on the edges represent the strength of the coupling influence.

[0028] For example, in this embodiment, step S126 may include the following sub-steps: Step S1261: Sort the component operation status data classified by mode in chronological order to generate component operation status time sequence data in each intelligent driving mode.

[0029] For component operating status data categorized by mode, the data is sorted based on timestamps. For example, in the urban road autonomous driving mode, the operating status data of all components in this mode are arranged in chronological order of collection time to form an operating status time series for each component. These time series are combined to generate component operating status time series data for each intelligent driving mode, with each time point in the time series data corresponding to a set of component operating status parameters.

[0030] Step S1262: Classify the fault pre-state change sequence data according to the intelligent driving mode, so that the fault pre-state change is matched with the corresponding driving mode, and generate fault pre-state data classified by mode.

[0031] We examine the intelligent driving mode information corresponding to each fault instance in the fault pre-fault state change sequence data, and then classify the above sequence data according to the intelligent driving mode. For example, fault pre-fault state change sequences occurring in the autonomous driving mode on urban roads are grouped into one category, generating fault pre-fault state data categorized by mode. In this way, each intelligent driving mode has a corresponding set of fault pre-fault state change sequences.

[0032] Step S1263: Align the time-series data of component operation status in each intelligent driving mode with the corresponding fault pre-fault status data classified by mode, so that the changes in component operation status and the changes in fault pre-fault status are accurately correlated in the time dimension, and generate patterned time-aligned data.

[0033] Taking the autonomous driving mode on urban roads as an example, the time-series data of component operating status in this mode is time-aligned with the corresponding fault pre-fault state data categorized by mode. The time point of fault occurrence is identified, and then the data from the component operating status time-series data corresponding to the same time period before the fault occurrence is matched with the fault pre-fault state change sequence data to ensure a one-to-one correspondence in the time dimension. After the above processing, patterned time-aligned data is generated, in which the changes in component operating status and the changes in fault pre-fault state are synchronized in time.

[0034] Step S1264: Extract component operating state change segments from patterned time-aligned data under each intelligent driving mode before the fault occurs, focusing on extracting the time difference and magnitude correlation of different component state changes, and generating patterned fault-related state change segment data.

[0035] From patterned time-aligned data, segments of component operational status changes in each intelligent driving mode prior to the occurrence of a fault are extracted. For example, in urban road autonomous driving mode, segments of component operational status data are extracted 10 minutes before each fault instance occurs. Then, the time difference between the state changes of different components in these segments is analyzed, i.e., the time interval between the start of a change in the state parameter of one component and the start of a change in the state parameter of another component; and the correlation of change magnitude, i.e., the proportional relationship between the change magnitudes of different component state parameters. This information is extracted to generate patterned fault-related state change segment data.

[0036] Step S1265: Combine the environment-component correlation data to analyze the mutual influence of the changes in the operating status of each component under different environmental conditions, eliminate non-coupling state changes caused by environmental factors, retain the direct coupling influence between components, and generate coupling change data after environmental exclusion.

[0037] Based on environment-component correlation data, the mutual influence of changes in the operating state of various components under different environmental conditions is analyzed. For example, in a high-temperature environment, the increase in the operating temperature of power system components may be caused by the ambient temperature, rather than by the coupling effect of other components. By using environment-component correlation data, these non-coupling state changes caused by environmental factors can be identified and excluded. Only those state changes caused by coupling effects resulting from direct interactions between components, unaffected by environmental factors, are retained, generating coupled change data after environmental exclusion.

[0038] Step S1266: Perform time-series alignment and comparison on the coupling change data after environmental exclusion, identify component pairs with sequential state changes, determine the initiator and receiver of the coupling influence, and generate state change time-series relationship data.

[0039] After excluding environmental factors, the coupling change data is time-series aligned to ensure consistency in the state change data of different components along the time axis. Then, by comparing the state change curves of different components, component pairs with a clear sequential state change are identified. For example, if the state parameter of component A changes first, followed by the state parameter of component B, then component A can be determined as the initiator of the coupling effect, and component B as the receiver. This information is recorded to generate state change time-series relationship data.

[0040] Step S1267: Calculate the correlation between changes in the same type of state parameters of different components, quantitatively evaluate the degree of correlation between a change in a certain state parameter of a component and the same state parameter changes of other components, and generate state change correlation data.

[0041] For similar state parameters, such as operating temperature, in the coupled change data after excluding environmental factors, the correlation between changes in this parameter and different components is calculated. Correlation analysis is used to quantify the degree of correlation between the operating temperature change of one component and the operating temperature changes of other components by calculating the correlation coefficient. The correlation coefficient ranges from -1 to 1; the closer the absolute value is to 1, the stronger the correlation. The calculated correlation coefficients and corresponding component pair information are then compiled to generate state change correlation data.

[0042] Step S1268: Based on the preset temporal stability threshold and correlation threshold, jointly determine the temporal relationship data of state change and the correlation data of state change, distinguish between accidental association and necessary coupling, and generate influence relationship judgment data.

[0043] For example, based on preset temporal stability and correlation thresholds, a joint judgment is made on temporal relationship data and correlation data of state changes to distinguish between accidental association and necessary coupling. The temporal stability threshold is set to 0.8. This threshold and correlation threshold are quantified by calculating the proportion of the same component pair whose state changes occur in the same order across at least 20 different fault instances. If this proportion is greater than or equal to 0.8, the temporal stability is deemed to meet the requirements. The correlation threshold is set to 0.6. This threshold is used to determine the correlation between components regarding changes in the same state parameter (such as operating temperature). The Pearson correlation coefficient is calculated; if the absolute value of the correlation coefficient is greater than or equal to 0.6, the two are considered significantly correlated. Only component pairs that simultaneously satisfy a temporal consistency ratio greater than or equal to 0.8 and an absolute value of the Pearson correlation coefficient greater than or equal to 0.6 are considered to have a necessary coupling relationship; otherwise, they are considered accidental associations. Based on the above joint judgment results, influence relationship judgment data is generated.

[0044] Step S1269: Determine the type of mutual influence relationship, distinguish whether it is mechanical coupling influence, hydraulic coupling influence, electrical coupling influence or signal coupling influence, extract information on the carrier and transmission mode of the coupling influence, and generate influence relationship type data.

[0045] For component pairs determined to have an inherent coupling relationship, the type of their mutual influence is further analyzed. By reviewing the vehicle's structural design drawings and system principle data, the connection methods and interaction mechanisms between components are determined. If the influence is generated through the transmission of force or motion via mechanical structures, it is a mechanical coupling influence; if the influence is generated through the transmission of pressure via hydraulic lines, it is a hydraulic coupling influence; if the influence is generated through the transmission of current or voltage via electrical lines, it is an electrical coupling influence; and if the influence is generated through the transmission of information via data signals, it is a signal coupling influence. Simultaneously, the carriers of the coupling influence, such as drive shafts in mechanical structures, oil pipes in hydraulic lines, wires in electrical lines, and communication protocols in signal interactions, as well as the transmission methods, such as force transmission, pressure transmission, current transmission, and signal transmission, are extracted to generate influence relationship type data.

[0046] Step S12610: Establish a mapping relationship model between the changes in state parameters of different components and generate influence parameter mapping data.

[0047] Based on the previously obtained data on the temporal relationship of state changes, the correlation of state changes, and the types of influence relationships, a mapping model between the changes in state parameters is established for each pair of components with an inherent coupling relationship. For example, for components A and B with mechanical coupling effects, a functional relationship model can be established between the vibration frequency changes of component A and component B. This functional relationship model can be linear or nonlinear, depending on the actual coupling characteristics. The above mapping relationship models and their parameter information are then compiled to generate influence parameter mapping data.

[0048] Step S12611: Associate and integrate the state change time sequence relationship data, state change correlation data, influence relationship judgment data, influence relationship type data, and influence parameter mapping data to generate multi-component coupling relationship data.

[0049] The state change time sequence data, state change correlation data, influence relationship judgment data, influence relationship type data, and influence parameter mapping data obtained in the above steps are correlated and integrated. Using component pairs as the basic unit, various types of data corresponding to each component pair are bound to form a complete component coupling relationship description. All these descriptions of component pairs are combined to generate multi-component coupling relationship data, which comprehensively reflects the coupling relationships between multiple components in the intelligent driving mode.

[0050] Step S127: Input the component operation status data, environment-component correlation data, fault pre-state change sequence data, and multi-component coupling relationship data classified by mode into the artificial intelligence modeling model. Through the artificial intelligence modeling model, learn the dynamic coupling law of multi-components under different intelligent driving modes and different environmental conditions, and generate an initial dynamic coupling model.

[0051] The AI ​​modeling system uses data on component operating status categorized by pattern, environment-component correlation data, pre-fault state change sequence data, and multi-component coupling relationship data as inputs. This AI modeling system employs a multi-input fusion deep neural network architecture, specifically configured as follows: The AI ​​modeling system comprises one input layer, five hidden layers, and one output layer. The input layer has 2048 nodes, corresponding to the total dimension of the input feature vector. The hidden layers are as follows: the first convolutional layer has a kernel size of 3, a stride of 1, and 64 filters; the second convolutional layer has a kernel size of 5, a stride of 1, and 128 filters; the third fully connected layer has 1024 neurons; the fourth fully connected layer has 512 neurons; and the fifth fully connected layer has 256 neurons. All hidden layers use the ReLU activation function, and a Dropout layer with a dropout rate of 0.3 is added after each fully connected layer. The model was trained using the Adam optimizer with an initial learning rate of 0.001, which decayed to 0.9 times the original rate every 10 training epochs. The loss function was mean squared error, the batch size was set to 64, the maximum number of training epochs was 200, and an early stopping method was employed (training stopped when the validation set loss did not decrease for 10 consecutive epochs). Through this configured deep neural network model, the model was able to learn the dynamic coupling patterns of multiple components under different intelligent driving modes and environmental conditions, thereby generating an initial dynamic coupling model containing all network layer weights and bias parameters.

[0052] Step S128: Adjust the parameters of the initial dynamic coupling model according to the control logic of the intelligent driving system, so that the adjusted dynamic coupling model can adapt to the interaction characteristics of components and control commands in the intelligent driving process. Integrate the adjusted dynamic coupling model with all related data to generate multi-component dynamic coupling modeling data in the intelligent driving scenario.

[0053] This multi-component dynamic coupling modeling data is a complete data package containing five core parts: an adjusted dynamic coupling model, a neural network structure description stored in JSON format, and pre-trained weight and bias parameter files; multi-component coupling relationship data, stored in a graph structure, where nodes represent components and edges represent coupling relationships; component operating status data categorized by mode, stored in tabular form and organized according to different driving modes, representing component state time series; environment-component association data, a dataset of association rules stored in IF-THEN format; and fault pre-fault state change sequence data, stored in a multi-dimensional array, representing pre-fault time series data. All data is packaged into a single HDF5 file, using different groups and datasets to organize the above five types of data, and includes a JSON index file to describe the relationships between the data groups.

[0054] The control logic of an intelligent driving system determines the interaction methods and response characteristics between components and control commands. For example, when an intelligent driving system issues an acceleration command, powertrain components need to adjust their output power accordingly, while braking system components may be in a standby state. Based on this control logic, the parameters of the initial dynamic coupling model are adjusted. For instance, a control command input layer is added, using control commands as one of the model's inputs, and the weight parameters related to control commands in the model are adjusted to ensure the model accurately reflects the components' responses to control commands. After adjustment, the adjusted dynamic coupling model is integrated with all associated data, including component operating status data categorized by mode, environment-component correlation data, fault pre-fault state change sequence data, and multi-component coupling relationship data, to form a complete dataset—the multi-component dynamic coupling modeling data for intelligent driving scenarios.

[0055] Step S130: Based on the dynamic coupling modeling data of multiple components in the intelligent driving scenario, analyze the chain transmission logic of faults among the core components of intelligent driving, and generate fault chain transmission path data of intelligent driving components.

[0056] In urban road autonomous driving scenarios, based on the dynamic coupling modeling data of multiple components in the intelligent driving scenario constructed above, the chain transmission logic of faults among core intelligent driving components is analyzed. Through in-depth analysis of the modeling data, it is clarified how faults are transmitted from one component to another, as well as the rules and characteristics in the transmission process, ultimately generating chain transmission path data of intelligent driving component faults.

[0057] In this embodiment, step S130 may include the following sub-steps: Step S131: Extract the adjusted dynamic coupling model, multi-component coupling relationship data, and fault pre-state change sequence data from the multi-component dynamic coupling modeling data in the intelligent driving scenario.

[0058] From the dynamic coupling modeling data of multiple components in intelligent driving scenarios, the adjusted dynamic coupling model, multi-component coupling relationship data, and fault pre-fault state change sequence data were selected. The adjusted dynamic coupling model contains the rules and parameters of dynamic coupling between components; the multi-component coupling relationship data describes the type and strength of coupling relationships between components; and the fault pre-fault state change sequence data records the state changes of components before the fault occurs.

[0059] Step S132: Based on the adjusted dynamic coupling model, analyze the transmission triggering conditions of different fault types among the core components of intelligent driving, and generate fault transmission triggering condition data.

[0060] The adjusted dynamic coupling model incorporates the coupling relationships and influence patterns between different components. For different fault types, such as mechanical, electrical, and signal faults, the conditions required for a fault to propagate from one component to another are determined by analyzing the threshold changes in the component state parameters and their mutual influence relationships. For example, for mechanical faults, fault propagation begins when the vibration amplitude of a component exceeds a certain threshold and the operating temperature of another component coupled with it reaches a specific value. These conditions are then compiled to generate fault propagation trigger condition data, with each trigger condition corresponding to a fault type and a pair of related components.

[0061] Step S133: Extract the connection link information of the core intelligent driving components by combining the multi-component coupling relationship data. The connection links of the core intelligent driving components include mechanical transmission links, hydraulic transmission links, electrical control links, and signal interaction links, and generate component connection link data.

[0062] The multi-component coupling relationship data contains the coupling relationship type and carrier information between components. Based on this information, the connection link information of core intelligent driving components is extracted. For mechanical coupling relationships, mechanical transmission links are identified, such as components connected via drive shafts, gears, etc.; for hydraulic coupling relationships, hydraulic transmission links are identified, such as components connected via oil pipes, hydraulic pumps, etc.; for electrical coupling relationships, electrical control links are identified, such as components connected via wires, relays, etc.; for signal coupling relationships, signal interaction links are identified, such as components exchanging signals via CAN bus, Ethernet, etc. This link information is then organized and structured to generate component connection link data.

[0063] In this embodiment, step S133 may include the following sub-steps: Step S1331: Extract the coupling influence relationship, influence mode, and influence carrier information between each intelligent driving core component from the multi-component coupling relationship data.

[0064] From the multi-component coupling relationship data, the coupling influence relationship between each pair of core intelligent driving components is extracted one by one, that is, which component affects which component; the influence method, such as whether it is a direct or indirect influence, whether it is a positive or negative influence, etc.; the influence carrier information, that is, through what physical or logical carrier the influence is realized, such as mechanical structure, hydraulic pipeline, electrical circuit, data signal, etc. The above information is recorded in the form of a triple, namely (source component, target component, influence method, influence carrier).

[0065] Step S1332: Based on the information of the influencing carrier, distinguish different types of connection links, classify the coupling influence relationship with mechanical structure as carrier as mechanical transmission link related data, and classify the coupling influence relationship with hydraulic pipeline as carrier as hydraulic transmission link related data.

[0066] The extracted information on the influencing carriers is analyzed, and different connection links are distinguished according to the type of carrier. If the influencing carrier is a mechanical structure, such as a drive shaft, connecting rod, or gear, the corresponding coupling influence relationship is classified as mechanical transmission link related data; if the influencing carrier is a hydraulic pipeline, such as an oil pipe or hydraulic hose, it is classified as hydraulic transmission link related data.

[0067] Step S1333: Classify the coupling influence relationship with electrical lines as the carrier as electrical control link related data, and classify the coupling influence relationship with data signals as signal interaction link related data.

[0068] Similarly, coupling effects on electrical circuits, such as wires and cables, are classified as electrical control link related data; coupling effects on data signals, such as CAN signals and Ethernet signals, are classified as signal interaction link related data.

[0069] Step S1334: Refine the data related to the mechanical transmission link, extract the component composition information, transmission ratio information, and connection information from the data related to the mechanical transmission link, form the composition and connection details of the mechanical transmission link, and generate detailed data of the mechanical transmission link.

[0070] For data related to the mechanical transmission link, further extract component composition information, i.e., which specific components are included in the link; transmission ratio information, such as the transmission ratio of gear transmission, belt transmission, etc.; and connection information, i.e., the connection points and connection methods between components. Record the above information in detail to form the composition and connection details of the mechanical transmission link, generating detailed mechanical transmission link data.

[0071] Step S1335: Refine the relevant data of the hydraulic transmission link, extract the hydraulic component composition information, hydraulic oil flow path information, and pressure transmission node information from the relevant data of the hydraulic transmission link, form the composition and flow details of the hydraulic transmission link, and generate detailed data of the hydraulic transmission link.

[0072] For data related to the hydraulic transmission link, extract information on the composition of hydraulic components, such as hydraulic pumps, hydraulic cylinders, and hydraulic valves; information on the flow path of hydraulic oil, i.e. the flow route of hydraulic oil from the pump to the actuator and back to the tank; and information on pressure transmission nodes, such as pressure monitoring points and pressure regulation nodes in the pipeline. The above information constitutes the composition and flow details of the hydraulic transmission link, generating detailed data on the hydraulic transmission link.

[0073] Step S1336: Refine the electrical control link related data, extract the electrical component composition information, current transmission path information, and control signal trigger node information from the electrical control link related data, form the composition and transmission details of the electrical control link, and generate detailed electrical control link data.

[0074] The detailed processing of electrical control link related data includes extracting information on the composition of electrical components, such as resistors, capacitors, inductors, transistors, integrated circuits, etc.; current transmission path information, that is, the flow path of current from the power source through each electrical component; and control signal triggering node information, such as nodes that generate control signals, such as switches and sensors. The above information forms the composition and transmission details of the electrical control link, generating detailed data of the electrical control link.

[0075] Step S1337: Refine the data related to the signal interaction link, extract the signal type information, signal transmission protocol information, and signal receiving and feedback node information from the data related to the signal interaction link, form the composition and interaction details of the signal interaction link, and generate detailed data of the signal interaction link.

[0076] The detailed processing of data related to the signal interaction link involves extracting signal type information, such as digital signals, analog signals, pulse signals, etc.; signal transmission protocol information, such as CAN protocol, LIN protocol, Ethernet protocol, etc.; and signal receiving and feedback node information, that is, the signal sender and receiver and the receiver's feedback mechanism. The above information constitutes the composition and interaction details of the signal interaction link, generating detailed data of the signal interaction link.

[0077] Step S1338: Integrate detailed data of mechanical transmission links, hydraulic transmission links, electrical control links, and signal interaction links to construct a complete connection link network for core components of intelligent driving.

[0078] The detailed data obtained above for mechanical transmission links, hydraulic transmission links, electrical control links, and signal interaction links are integrated. Using core components of intelligent driving as nodes and various links as edges, a complete connection link network is constructed. In this network, each node represents a core component, and each edge represents a connection link, with edge attributes including link type, composition details, and other information.

[0079] Step S1339: Mark the intersection nodes of each link in the link network, that is, the components or connection parts shared by different types of links, and generate link intersection node data.

[0080] In the constructed network of interconnected links, identify the components or connection points shared by different types of links; these points are the link intersection nodes. For example, if a component belongs to both a mechanical transmission link and an electrical control link, then that component is the intersection node of these two types of links. Label these intersection nodes and their respective link types to generate link intersection node data.

[0081] Step S13310: Associate the complete connection link network with the link intersection node data to generate component connection link data containing detailed composition, connection relationship and intersection node of each link.

[0082] By associating the complete link network with the data of the link intersection nodes, each intersection node in the link network can be mapped to its relevant link information. In this way, the generated component connection link data not only includes the detailed composition and connection relationship of each link, but also clarifies the location and attributes of the link intersection nodes.

[0083] Step S134: Extract the state features of the fault-initiating component from the fault pre-fault state change sequence data, determine the component where the fault first occurred and the initial fault state, and generate fault initiation source data.

[0084] Analyze the sequence data of pre-fault state changes to observe the changes in the state parameters of each component before the fault occurs. Typically, the component where the fault first occurs will show the first abnormal changes in its state parameters. By comparing the state parameter ranges of components under normal conditions, identify those components whose state parameters exceed the normal range and determine their initial fault state, such as excessively high operating temperature or abnormal vibration. Record this information to generate fault initiation source data, which includes the identifier of the fault-initiating component and a description of the initial fault state.

[0085] Step S135: Based on the component connection link data and fault transmission trigger condition data, deduce the order in which the fault is transmitted from the originating component to other related components based on preset logical rules, extract the transmission order information of the fault in different links, and generate fault link transmission order data.

[0086] Component connection link data provides the connection relationships between components, while fault transmission trigger condition data specifies the conditions for fault transmission. Based on preset logical rules, such as "faults are prioritized for transmission along directly connected links" and "links that meet the trigger conditions are prioritized for fault transmission," starting from the fault originating component, each link connected to it is sequentially checked to see if it meets the fault transmission trigger conditions. For links that meet the conditions, the order in which the fault is transmitted to the next component is determined. Simultaneously, the transmission order of the fault in different types of links is recorded, such as first transmitting in mechanical transmission links, then in signal interaction links, etc. This transmission order information is then compiled to generate fault link transmission order data.

[0087] For example, based on component connection link data and fault transmission triggering condition data, a predefined set of logical rules is used to deduce the order in which a fault is transmitted from the originating component to other related components. This set of logical rules specifically includes: Rule 1, Priority Transmission Rule: If the fault originating component A has both mechanical connection links and signal interaction links with components B and C, the fault is preferentially transmitted along the mechanical transmission link until the transmission conditions of that link are no longer met, then it is transmitted along the signal interaction link; Rule 2, Triggering Condition Transmission Rule: If the state parameters (such as operating temperature) of the fault originating component A reach or exceed its preset fault transmission triggering threshold, and the related component B meets the conditions in the fault transmission triggering condition data, the fault is transmitted from A to B; Rule 3, Directional Transmission Rule: If the fault transmission triggering condition data indicates that fault transmission is unidirectional, the fault transmission direction is from the initiator to the receiver, and reverse transmission is not allowed. Based on the above rule set, starting from the fault originating component, all its connection links are traversed sequentially to determine whether each link meets the corresponding triggering conditions, thereby determining the order in which the fault is transmitted to the next component. At the same time, the transmission order of the fault in different types of links is recorded to generate fault link transmission order data.

[0088] Step S136: Track the state changes of the fault during the transmission process between components, record the state parameter changes and control signal response changes of each component during the fault transmission process, and generate fault transmission state change data.

[0089] During fault propagation, the changes in the state parameters and control signal responses of each component are tracked in real time. For example, when a fault propagates from the originating component to the next component, changes in the component's operating temperature, operating pressure, vibration, and other state parameters are recorded, as well as changes in the response of the intelligent driving system to the control signals emitted by the component, such as changes in the strength and frequency of the control signals. These changes are recorded in chronological order and by component number to generate fault propagation state change data.

[0090] Step S137: Mark the key transmission nodes in the fault transmission process and generate fault key transmission node data. The key transmission nodes include fault cross-link transmission nodes, fault aggravation nodes, and fault-induced multi-component collaborative anomaly nodes.

[0091] In the fault propagation path, key propagation nodes are identified and marked. Cross-link fault propagation nodes are nodes where a fault is propagated from one type of link to another; fault aggravation nodes are nodes where the fault causes a significant increase in the degree of abnormality in state parameters; fault-induced multi-component collaborative anomaly nodes are nodes where, after the fault propagates to this node, multiple related components simultaneously exhibit abnormalities. The locations, link types, and corresponding fault characteristics of these key propagation nodes are recorded to generate fault key propagation node data.

[0092] Step S138: Connect the fault origin data, fault link transmission order data, and fault key transmission node data in sequence according to the fault transmission time to construct a preliminary fault chain transmission path.

[0093] Following the chronological order of fault propagation, the fault origin data, fault link propagation sequence data, and fault critical propagation node data are sequentially linked together. Starting from the fault origin component, each propagation node and critical propagation node are connected sequentially according to the fault link propagation sequence, forming a complete fault propagation path, i.e., a preliminary fault chain propagation path. This real-time data fusion path reflects the entire process from fault occurrence to propagation.

[0094] Step S139: Adjust the initial fault chain transmission path by combining the power transmission logic and control signal interaction logic of the intelligent driving system, so that the adjusted fault chain transmission path conforms to the actual collaborative working logic of the components in the intelligent driving process, and generate a collaboratively adapted fault chain transmission path.

[0095] The power transmission logic of an intelligent driving system determines how power is transmitted between components, while the control signal interaction logic determines the signal communication rules between components. Based on these logics, the initial fault chain propagation path is adjusted. For example, if, according to the power transmission logic, a fault should first propagate to a component in the power system and then to other components, but the initial path reverses this order, adjustment is necessary. Through these adjustments, the fault chain propagation path is made consistent with the actual collaborative working logic of the components, generating a collaboratively adapted fault chain propagation path.

[0096] Step S1310: Extract the transmission feature information in the collaborative adaptation fault chain transmission path. The transmission feature information includes the transmission duration of the fault in each link, the rate of state change during the fault transmission process, and the control signal delay time during fault transmission, and generate fault chain transmission feature data.

[0097] From the collaborative adaptation fault chain propagation path, propagation feature information is extracted. For each link, the propagation time of the fault in that link is calculated, that is, the time elapsed from the fault entering the link to leaving the link; the state change rate during the fault propagation process, that is, how quickly the component state parameters change from normal to abnormal; and the control signal delay time during fault propagation, that is, the time difference between the control signal being issued and the component responding. After organizing the above feature information, fault chain propagation feature data is generated.

[0098] Step S1311: Associate and integrate the collaborative adaptation fault chain transmission path with the fault chain transmission feature data to generate intelligent driving component fault chain transmission path data.

[0099] By associating the collaboratively adapted fault chain propagation path with fault chain propagation characteristic data, each node and link in the path corresponds to specific propagation characteristic information. For example, on a mechanical transmission link in the path, characteristic data such as its propagation duration, state change rate, and control signal delay time are associated. In this way, the generated intelligent driving component fault chain propagation path data includes both the path information of fault propagation and the characteristic parameters during the propagation process.

[0100] Step S140: Integrate real-time environmental data of intelligent driving with real-time operating status data of multiple components, and dynamically correlate it with fault chain transmission path data of intelligent driving components to generate real-time coupled status data of multiple intelligent driving components.

[0101] In urban road autonomous driving scenarios, the intelligent driving system collects environmental data and component operating status data in real time, integrates these two types of data, and dynamically associates them with the fault chain transmission path data of intelligent driving components to generate real-time coupling status data of multiple intelligent driving components. This real-time coupling status data of multiple intelligent driving components can reflect the current coupling status of the components and their association with the fault transmission path in real time.

[0102] In this embodiment, step S140 may include the following sub-steps: Step S141: Real-time environmental data of intelligent driving is collected in real time by the environmental perception module of the intelligent driving system. The real-time environmental data of intelligent driving includes real-time road type data, real-time traffic flow data, real-time weather condition data, real-time lighting condition data, and real-time obstacle distribution data.

[0103] The environmental perception module of the intelligent driving system operates continuously, collecting real-time environmental data through various sensors. Real-time road type data is determined by road signs captured by cameras and map data, such as whether it is an urban expressway, a regular highway, or a residential road; real-time traffic flow data is obtained by detecting the number and density of surrounding vehicles and pedestrians through radar and cameras; real-time weather condition data is collected by onboard weather sensors, including real-time temperature, humidity, and precipitation; real-time lighting condition data is obtained by lighting sensors, such as current light intensity and whether there is direct sunlight; real-time obstacle distribution data is obtained by scanning the surrounding environment through lidar and cameras, including the location, size, direction of movement, and speed of obstacles. The above real-time environmental data is continuously updated at certain time intervals and stored in a structured data format.

[0104] Step S142: The component status monitoring module of the intelligent driving system collects real-time operating status data of multiple components. The real-time operating status data of multiple components includes real-time component data of the steering system, the braking system, the power system, the perception system, and the decision control system. The real-time component data of each system includes real-time operating temperature data, real-time operating pressure data, real-time vibration data, real-time energy consumption data, real-time running time data, and real-time control signal interaction data.

[0105] The component status monitoring module also collects real-time operating status data of multiple components. For the steering system, it collects real-time data on the steering gear's operating temperature, power steering pump's operating pressure, steering motor vibration, energy consumption, cumulative runtime, and steering control signal interaction. For the braking system, it collects data on brake pad temperature, brake fluid pressure, brake disc vibration, braking system energy consumption, runtime, and braking control signal interaction. The real-time operating status data collection methods for the power system, sensing system, and decision control system are similar, collecting the aforementioned real-time data from key components within their respective systems. This data is also stored in time series format to ensure synchronization with environmental data.

[0106] Step S143: Input the real-time environmental data of intelligent driving and the real-time operating status data of multiple components into the real-time data fusion module of the artificial intelligence model, perform data time-series alignment processing, so that the environmental data and the operating status data of components correspond accurately in the time dimension, and generate time-series aligned real-time fused data.

[0107] The collected real-time environmental data and real-time operational status data of multiple components for intelligent driving are input into the real-time data fusion module of the artificial intelligence model. This module first timestamps the two types of data, matching environmental data and component operational status data at the same moment by comparing the data collection times. For data with incompletely consistent timestamps, interpolation is used to ensure that there is corresponding environmental data and component operational status data at every time point. After this time-series alignment process, time-aligned real-time fused data is generated, in which environmental data and component operational status data correspond one-to-one in the time dimension.

[0108] Step S144: Extract the real-time operating status features of each core intelligent driving component from the time-aligned real-time fused data to generate real-time status feature data for a single component.

[0109] Feature extraction is performed on the time-aligned real-time fused data. For each core component of intelligent driving, features reflecting its current operating state are extracted from its operational status data. For example, for the engine of the power system, the mean, variance, and rate of change of its operating temperature, the peak and trough values ​​of its operating pressure, and the frequency characteristics of its vibration signal are extracted. These features are represented in vector form, with each vector corresponding to the real-time operating status feature of a component. The feature vectors of all components are combined to generate real-time status feature data for a single component.

[0110] Step S145: Extract collaborative adaptation fault chain transmission path and fault chain transmission feature data from the fault chain transmission path data of intelligent driving components.

[0111] From the previously generated fault chain propagation path data of intelligent driving components, we extract the collaborative adaptation fault chain propagation path and fault chain propagation feature data. The collaborative adaptation fault chain propagation path describes the path structure of fault propagation between components, while the fault chain propagation feature data includes characteristic parameters such as propagation duration and state change rate.

[0112] Step S146: Analyze the correlation between the real-time status feature data of each individual component, mine the dynamic coupling information between each component under the current intelligent driving condition, and generate real-time dynamic coupling data.

[0113] In urban road autonomous driving scenarios, the correlation between real-time state feature data of individual components is analyzed to uncover information on the dynamic coupling between components under current operating conditions. Through the analysis of feature data, it becomes clear how components influence and interact with each other.

[0114] In this embodiment, step S146 may include the following sub-steps: Step S1461: Extract real-time operating temperature features, real-time operating pressure features, real-time vibration features, real-time energy consumption features, and real-time control signal interaction features from the real-time status feature data of each individual component, and generate a multi-dimensional real-time status feature set for each component.

[0115] For each component's real-time status feature data, we extract real-time operating temperature features, such as the current temperature value and its trend; real-time operating pressure features, such as the pressure magnitude and fluctuation range; real-time vibration features, such as the vibration frequency and amplitude; real-time energy consumption features, such as energy consumption per unit time and energy consumption change rate; and real-time control signal interaction features, such as the signal transmission frequency and signal strength. These features are then combined to form a multi-dimensional real-time status feature set for each component.

[0116] Step S1462: Select the core components of the intelligent driving decision control system as the benchmark components, extract the multi-dimensional real-time state feature set of the benchmark components, and generate the benchmark feature set.

[0117] Among numerous core components of intelligent driving, core components of the decision-making and control system, such as the central control unit, are selected as benchmark components. A multi-dimensional real-time state feature set of this benchmark component is extracted, including features such as operating temperature, operating pressure, vibration, energy consumption, and control signal interaction, to generate a benchmark feature set.

[0118] Step S1463: Compare the multi-dimensional real-time state feature sets of other core intelligent driving components with the benchmark feature set one by one, analyze the synchronicity of changes in the same dimension features between different components, focus on the synchronous response of control signal interaction features, and generate feature synchronization change data.

[0119] The multi-dimensional real-time state feature sets of other core intelligent driving components, excluding the benchmark component, are compared one by one with the benchmark feature set. For each identical feature dimension, such as operating temperature, the synchronous changes are analyzed; that is, when the feature value of one component rises or falls, does the corresponding feature value of another component also rise or fall accordingly, and what is the time difference between the changes? Special attention is paid to the synchronous response of control signal interaction features; that is, after the benchmark component sends a control signal, are the times when other components receive and respond to the signal within a reasonable range, and does the intensity of the response match the signal? The results of the above synchronization analysis are recorded to generate feature synchronization change data.

[0120] Step S1464: Calculate the degree of mutual influence between the reference component and other components in the same dimension, determine the strength of the effect of a change in a certain dimension of the reference component on the same dimension of other components, especially the strength of the influence of the change in the control signal on the control signal response characteristics of other components, and generate feature influence strength data.

[0121] For the same dimensional characteristics of the reference component and other components, the degree of mutual influence between them is calculated. Correlation analysis is used to determine the intensity of the influence by calculating the ratio or proportion between the changes in characteristics. For example, the ratio of the change in the control signal interaction characteristic of the reference component to the change in the control signal response characteristic of other components reflects the intensity of the influence of the control signal change on the response characteristic. The calculated influence intensity values ​​are then organized to generate characteristic influence intensity data.

[0122] Step S1465: Replace the reference component with the core component of the intelligent driving power system, repeat the above process of reference feature selection, feature comparison, synchronization analysis, and influence intensity calculation, and generate a second set of feature synchronization change data and feature influence intensity data.

[0123] The benchmark component is replaced with a core component of the intelligent driving power system, such as an engine control unit. Following the methods in steps S1462 to S1464, a multi-dimensional real-time state feature set of the benchmark component is extracted, compared with the feature sets of other components, the synchronicity is analyzed, the influence intensity is calculated, and a second set of feature synchronization change data and feature influence intensity data is generated.

[0124] Step S1466: Replace the reference component with the core component of the intelligent driving steering system again, repeat the above process, and generate the third set of feature synchronous change data and feature influence intensity data.

[0125] The benchmark components were then replaced with core components of the intelligent driving steering system, such as the steering control unit. The same feature selection, comparison, analysis, and calculation process was performed to generate a third set of feature synchronization change data and feature influence intensity data.

[0126] Step S1467: Integrate three sets of feature synchronization change data and feature influence intensity data to construct a multi-dimensional feature coupling relationship network between various intelligent driving core components, generate feature coupling network data, and select coupling relationships that meet the conditions from the feature coupling network data according to the preset coupling strength threshold and synchronization threshold to generate active coupling relationship data.

[0127] The three sets of feature synchronization change data and feature influence intensity data obtained earlier are integrated. Using core components of intelligent driving as nodes and feature synchronization change and influence intensity as edge attributes, a multi-dimensional feature coupling relationship network is constructed. In this network, the edge weights are jointly determined by influence intensity and synchronization. Then, based on preset coupling strength thresholds and synchronization thresholds, coupling relationships with influence intensity greater than the coupling strength threshold and synchronization higher than the synchronization threshold are selected. These coupling relationships are considered active coupling relationships under the current operating condition, generating active coupling relationship data.

[0128] Step S1468: Analyze the direction of each coupling relationship in the active coupling relationship data, determine whether it is a unidirectional influence or a bidirectional mutual influence, extract the information of the initiating component and the affected component, and generate coupling direction data.

[0129] For each coupling relationship in the active coupling relationship data, analyze its direction of action. If only one component affects another without any reverse influence, it is a unidirectional influence; if the two components influence each other, it is a bidirectional mutual influence. Simultaneously, determine the identification information of the initiating and affected components to generate coupling direction data.

[0130] Step S1469: Match the active coupling relationship data with the pre-stored coupling relationship mechanism knowledge base to generate coupling mechanism description data.

[0131] The pre-stored coupling mechanism knowledge base contains various known component coupling relationships and their mechanisms of action. Each coupling relationship in the active coupling relationship data is matched with an entry in the knowledge base to find the corresponding coupling mechanism description, such as force transmission due to mechanical connections or the control effect of electrical signals. These mechanism descriptions are then extracted to generate coupling mechanism description data.

[0132] Step S14610: Combine and integrate the active coupling relationship data, coupling direction data, and coupling mechanism description data to generate real-time dynamic coupling data.

[0133] By associating active coupling relationship data, coupling direction data, and coupling mechanism description data, each active coupling relationship is accompanied by an action direction and mechanism description. In this way, the generated real-time dynamic coupling data comprehensively describes the dynamic coupling between various components under the current intelligent driving conditions.

[0134] Step S147: Compare the real-time status feature data of a single component with the fault transmission status change data of the corresponding component in the collaborative adaptation fault chain transmission path, locate the corresponding position of the current operating status of each component in the fault chain transmission path, and generate path position matching data.

[0135] For each component, its real-time status feature data is compared with the fault transmission status change data of that component in the collaborative adaptation fault transmission chain path. By comparing the similarity of feature values, it is determined which stage in the fault transmission path the current operating state of the component is closest to, thus locating its corresponding position in the fault transmission chain path. The position information of all components is recorded to generate path position matching data.

[0136] Step S148: Combining the transmission duration, state change rate, and control signal delay time in the fault chain transmission characteristic data, analyze the trend of the current operating state of each component evolving into a fault state, and generate state evolution trend data.

[0137] Using parameters such as transmission duration, state change rate, and control signal delay time from the fault chain propagation characteristic data, combined with the current real-time state characteristic data of components, the following specific mathematical model is used to analyze the trend of each component evolving towards a fault state. For fault arrival time prediction, let the current state parameter value of component i be Scurr(i), its fault trigger threshold be Sfault(i), and its current state change rate be vchange(i), which is calculated using the linear regression slope of the most recent 10 sampling points. Then, the predicted time Ttofault(i) required for the component to reach a fault state is equal to the difference between Sfault(i) and Scurr(i), divided by vchange(i). This calculation is valid when vchange(i) is greater than 0 and Scurr(i) is less than Sfault(i). For fault propagation probability judgment, let the current delay time of the control signal be Tdelaycurr, and its historical normal fluctuation upper limit be Tdelaynormalmax. The delay time deviation Ddelay is calculated as Tdelaycurr divided by Tdelaynormalmax. If Ddelay is greater than 1.2, it is determined that the control signal response has significantly abnormalized, and the probability of fault propagation increases. By combining Ttofault(i) and Ddelay, quantitative state evolution trend data is generated.

[0138] Step S149: Integrate real-time dynamic coupling data, path location matching data, and state evolution trend data to form preliminary real-time coupling state data.

[0139] Real-time dynamic coupling data, path location matching data, and state evolution trend data are integrated. The dynamic coupling information of each component, its position information in the fault path, and its state evolution trend information are correlated to form preliminary real-time coupling state data. This preliminary real-time coupling state data reflects the current coupling state between components and its relationship with the fault transmission path.

[0140] Step S1410: Supplement the current control mode information, power output status information, and path planning information of the intelligent driving system into the preliminary real-time coupling status data, generate the real-time coupling status data after control supplementation, and perform structured integration of the real-time coupling status data after control supplementation to generate real-time coupling status data of multiple intelligent driving components.

[0141] The system acquires current control mode information, such as the level of autonomous driving and driving mode selection; power output status information, such as engine output power and motor speed; and path planning information, such as the current route, destination, and estimated travel time. This information is then added to the initial real-time coupling state data to generate supplemented real-time coupling state data. This data is then structured and integrated according to a specific format and organization to generate real-time coupling state data for multiple intelligent driving components. This multi-component real-time coupling state data comprehensively and accurately reflects the current real-time coupling state of the various intelligent driving components.

[0142] Step S150: Based on the real-time coupled status data of multiple intelligent driving components and the fault chain transmission path data of intelligent driving components, the evolution process of faults under intelligent driving conditions is deduced, and the fault prediction results of intelligent driving components are generated.

[0143] In urban autonomous driving scenarios, the evolution of faults under current operating conditions is simulated based on real-time coupled state data of multiple intelligent driving components and fault chain propagation path data of these components. By analyzing the real-time coupling state of components and fault propagation paths, the development direction, scope, and possible timing of faults are predicted, ultimately generating fault prediction results for intelligent driving components.

[0144] In this embodiment, step S150 may include the following sub-steps: Step S151: Extract real-time dynamic coupling data, path position matching data, state evolution trend data, control mode information, power output state information, and path planning information from the real-time coupling state data of multiple intelligent driving components.

[0145] From the real-time coupling status data of multiple components in intelligent driving, the following data are extracted: real-time dynamic coupling data, which describes the dynamic coupling relationship between components; path position matching data, which indicates the position of each component in the fault transmission path; state evolution trend data, which reflects the trend of components evolving towards a fault state; control mode information, which describes the current control mode of the intelligent driving system; power output status information, which shows the output of the power system; and path planning information, which provides the current driving route planning.

[0146] Step S152: Extract the collaborative adaptation fault chain transmission path, fault chain transmission feature data, and fault key transmission node data from the intelligent driving component fault chain transmission path data.

[0147] Similarly, collaborative adaptation fault chain transmission paths are extracted from the fault chain transmission path data of intelligent driving components. These collaborative adaptation fault chain transmission paths are the specific paths of fault transmission. Fault chain transmission feature data includes features such as transmission duration and state change rate. Fault key transmission node data marks the key nodes in the fault transmission process.

[0148] Step S153: Determine the fault chain transmission stage corresponding to the current multi-component coupled operation state based on the path location matching data, and generate the current transmission stage data.

[0149] Based on the location information of each component in the fault chain transmission path from the path location matching data, a comprehensive determination is made as to which stage of the fault chain transmission path the current coupled operation state of multiple components is in. For example, if most components are in the initial stage of fault transmission, then the current transmission stage is the initial stage; if some components have already entered the intermediate stage of fault transmission, then the current transmission stage is the intermediate stage. The determined current transmission stage is recorded to generate current transmission stage data.

[0150] Step S154: Combining state evolution trend data and fault chain transmission characteristic data, deduce the subsequent fault transmission process in the current transmission stage, predict the next component to which the fault will be transmitted, the time required for transmission, and the changes in control signals during the transmission process, and generate fault transmission prediction data.

[0151] In urban road autonomous driving scenarios, by combining state evolution trend data and fault chain propagation characteristic data, the subsequent fault propagation process in the current propagation stage is extrapolated. By analyzing the state evolution trends and fault propagation characteristics of components, the next stage of fault development is predicted.

[0152] In this embodiment, step S154 may include the following sub-steps: Step S1541: Extract the rate of change, direction of change, and trend of change of control signal response of each component from the current state to the fault state from the state evolution trend data, and generate state evolution parameter data.

[0153] From the state evolution trend data, extract the rate of change of each component's current state towards the fault state, i.e., how fast the state parameters change over time; the direction of evolution, i.e., whether the state parameters change towards or away from the fault state; and the trend of control signal response, i.e., how the response time and intensity of the control signal change over time. Integrate these parameters to generate state evolution parameter data.

[0154] Step S1542: Extract the fault transmission duration, state change rate, control signal delay time, and fault transmission direction from the fault chain transmission feature data to generate fault transmission parameter data.

[0155] From the fault chain propagation characteristic data, we extract the following parameters: the transmission time of the fault in different links (i.e., the time required for the fault to travel from one component to another on a certain link); the state change rate (i.e., the speed at which the state parameters of the components change during the fault propagation); the control signal delay time (i.e., the delay of the control signal during the propagation); and the fault propagation direction (i.e., whether the fault propagates unidirectionally or bidirectionally on the link). These parameters are then extracted to generate fault propagation parameter data.

[0156] Step S1543: Determine the component where the current fault is located, the current fault status parameters, the current fault transmission progress, and the current link type from the current transmission stage data, and generate the current fault status data.

[0157] Based on the data from the current transmission stage, identify the specific component where the current fault occurs, along with its current fault status parameters, such as operating temperature and pressure; the current fault transmission progress, i.e., the degree of fault development on the current component; and the current link type, such as a mechanical transmission link or a signal interaction link. After organizing the above information, generate the current fault status data.

[0158] Step S1544: Based on the current fault status data and fault transmission parameter data, determine the transmission conditions for the fault to be transmitted from the current component to the next related component, extract the threshold values ​​of the status parameters and the threshold values ​​of the control signal changes required for transmission, and generate fault transmission condition data.

[0159] By combining current fault status data and fault propagation parameter data, we analyze the conditions that must be met for a fault to propagate from the current component to the next related component. For example, the operating temperature of the current component needs to reach a certain threshold, and the change in the control signal needs to exceed a certain threshold. We extract these status parameter thresholds and control signal change thresholds to generate fault propagation condition data.

[0160] Step S1545: Analyze the rate of change of the current component's state and the trend of the control signal response in the state evolution parameter data, determine whether the current component's fault state meets the fault propagation conditions, and generate data on the fulfillment of propagation conditions.

[0161] Based on the rate of change of the current component's state and the trend of the control signal response in the state evolution parameter data, predict whether the component's state parameters and control signals will reach the threshold set in the fault propagation condition data within a certain period of time. If the prediction result indicates that it will reach the threshold, the current fault state meets the propagation condition; otherwise, it does not. Record the judgment result to generate propagation condition satisfaction data.

[0162] Step S1546: If the transmission conditions are met, combine the transmission duration in the fault transmission parameter data and the coupling influence strength from the current component to the next component to calculate the time required for fault transmission. At the same time, predict the delay time and signal strength change of the control signal during the transmission process, and generate transmission time calculation data and control signal change prediction data.

[0163] If the fault transmission condition data indicates that the current fault state meets the transmission conditions, then the time required for fault transmission is calculated by combining the transmission duration in the fault transmission parameter data and the coupling influence strength from the current component to the next component. The stronger the coupling influence, the shorter the transmission duration may be. Simultaneously, based on the control signal delay time and current control signal response change trend in the fault transmission parameter data, the changes in control signal delay time and signal strength during fault transmission are predicted, generating transmission time calculation data and control signal change prediction data.

[0164] Step S1547: Based on the fault transmission direction in the fault transmission parameter data and the correlation in the component connection link data, determine the next component to which the fault will be transmitted, extract the link type information of the next component, and generate the data of the next faulty component.

[0165] Based on the fault propagation direction in the fault propagation parameter data and the association between the current component and other components in the component connection link data, the next component to which the fault will be propagated is determined. Simultaneously, the link type information of the next component is extracted, such as mechanical transmission link, hydraulic transmission link, etc., to generate the data for the next faulty component.

[0166] Step S1548: Predict the initial fault state of the next component after receiving the fault transmission, and generate the initial fault state data of the next component by combining the state change rate in the fault transmission parameter data, the current fault state parameters, and the control signal change prediction data.

[0167] Based on the state change rate, current fault state parameters, and control signal change prediction data in the fault transmission parameter data, the initial fault state of the next component after receiving the fault transmission is predicted. For example, if the current component's operating temperature is T, the state change rate is v, and the transmission time is t, then the initial operating temperature of the next component might be T + v * t (adjusted to account for factors such as coupling strength). The predicted initial fault state parameters are recorded to generate the initial fault state data for the next component.

[0168] Step S1549: If the transmission conditions are not met, analyze the difference between the current component state evolution parameter data and the fault transmission condition data, predict the time required to meet the transmission conditions, and analyze the possible control signal adjustment trend to meet the transmission conditions, and generate transmission condition satisfaction time data and control signal adjustment trend data.

[0169] If the transmission conditions are not met, analyze the differences between the current component state evolution parameter data (state change rate, control signal response change trend, etc.) and the threshold values ​​in the fault transmission condition data. Based on these differences and the state change rate, predict how long it will take for the component state to meet the transmission conditions. Simultaneously, analyze what adjustments the intelligent driving system might make to the control signals to accelerate reaching the transmission conditions, such as increasing signal strength or shortening the signal period, generating transmission condition fulfillment time data and control signal adjustment trend data.

[0170] Step S15410: Combine and integrate the data of the next faulty component, the transmission time calculation data, the control signal change prediction data, the initial fault state data of the next component, the transmission condition satisfaction time data, and the control signal adjustment trend data to generate fault transmission prediction data.

[0171] The data obtained from the above steps—including the next faulty component data, transmission time calculation data, control signal change prediction data, initial fault state data of the next component, transmission condition fulfillment time data, and control signal adjustment trend data—are correlated and integrated. Following the logical order of fault transmission, this data is combined to form fault transmission prediction data, which comprehensively predicts the subsequent transmission of the fault in the current stage.

[0172] Step S155: Based on real-time dynamic coupling data and control mode information, analyze the impact of the coupling effect of each component on the fault evolution during the fault transmission process, combine the path planning information of intelligent driving to judge the trend of the impact of fault evolution on driving safety, and generate fault evolution trend judgment data.

[0173] Real-time dynamic coupling data describes the mutual influence between components, while control mode information determines the system's control strategy. Combining these two, this study analyzes how the coupling between components affects fault evolution during fault propagation, such as whether it accelerates or slows down fault development. Simultaneously, by incorporating intelligent driving path planning information, such as the current road type, whether there are curves ahead, and school zones, the study assesses the potential impact of fault evolution on driving safety, such as whether it will lead to vehicle loss of control or increased braking distance. The results of these analyses are then compiled into fault evolution trend judgment data.

[0174] Step S156: Based on the fault critical transmission node data, predict the arrival time, corresponding status, and multi-component collaborative anomalies of each critical node in the subsequent transmission process of the fault, and generate critical node prediction data.

[0175] The critical fault propagation node data marks the key nodes in the fault propagation process. Based on fault propagation prediction data and fault chain propagation characteristic data, the time for the fault to reach each critical node is predicted. Simultaneously, the state parameters of the critical node at the time of fault arrival and potential multi-component coordinated anomalies are predicted, such as simultaneous vibration anomalies in multiple components or control signal delays. This predicted information is recorded to generate critical node prediction data.

[0176] Step S157: Combining fault transmission prediction data, fault evolution trend judgment data, and key node prediction data, determine the range of components that may ultimately be involved in the fault, the final time of the fault occurrence, and the impact of the fault on the switching of intelligent driving modes, and generate fault final state prediction data.

[0177] By combining the fault propagation path and time from the fault propagation prediction data, the impact trend on driving safety from the fault evolution trend judgment data, and the situation of key nodes from the key node prediction data, we can determine the range of components that may ultimately be affected by the fault, i.e., which components will be affected by the fault; the final moment of the fault occurrence, i.e., when the fault develops to its most severe stage; and the impact of the fault on the switching of intelligent driving modes, such as whether it is necessary to switch from autonomous driving mode to manual driving mode, or to a degraded autonomous driving mode. The determined information is then used to generate the final fault state prediction data.

[0178] Step S158: Extract the core prediction information from the final state prediction data of the fault, including the list of parts involved in the fault, the time of the fault occurrence, the division of the fault evolution stages, and the scope of intelligent driving functions affected by the fault, and generate core prediction data.

[0179] Core predictive information is extracted from the final state prediction data of the fault. The fault-affected component list details all components potentially affected by the fault; the fault occurrence time clarifies the final time of the fault's occurrence; the fault evolution stage division divides the fault's development process into different stages; and the scope of intelligent driving functions affected by the fault indicates which intelligent driving functions, such as automatic steering and automatic braking, will be affected. This core information is then compiled into core predictive data.

[0180] Step S159: Associate the core prediction data with the corresponding real-time intelligent driving environment data and real-time operating status data of multiple components, label the intelligent driving working condition background corresponding to the prediction results, generate associated labeled prediction data, standardize and organize the associated labeled prediction data, and generate intelligent driving component fault prediction results that include fault type, involved component links, occurrence time, evolution process, and scope of affected functions.

[0181] The core prediction data is correlated with corresponding real-time intelligent driving environment data and real-time operating status data of multiple components, so that the prediction results can be mapped to specific driving conditions, such as fault prediction in urban congestion or when a component's real-time operating temperature is too high. Then, the correlated and labeled prediction data is standardized and organized, unifying the data format and description method, and finally generating intelligent driving component fault prediction results that include fault type (such as mechanical fault, electrical fault, etc.), involved component links (fault transmission path), occurrence time, evolution process (development status at each stage), and scope of affected functions.

[0182] Furthermore, after step S150, the method may further include: step S160: using the correspondence between real-time operating status data of multiple intelligent driving components and fault prediction results of intelligent driving components, iteratively updating the dynamic coupling modeling data of multiple components in the intelligent driving scenario, so as to realize the dynamic adaptation of the artificial intelligence model to the intelligent driving working condition.

[0183] In urban road autonomous driving scenarios, after a period of operation, a large amount of real-time operational status data of multiple intelligent driving components and corresponding fault prediction results of these components have been accumulated. Utilizing the correspondences between these data, the dynamic coupling modeling data of multiple components in intelligent driving scenarios is iteratively updated, enabling the artificial intelligence model to better adapt to constantly changing intelligent driving conditions.

[0184] Step S161: Collect real-time operating status data of multiple intelligent driving components and corresponding intelligent driving component fault prediction results within a preset time period, and generate a real-time data and prediction result dataset.

[0185] Set a preset time period, such as one month. Within this period, continuously collect real-time operating status data of multiple intelligent driving components and corresponding fault prediction results for these components. Organize the above data in chronological order, with each set of real-time operating status data corresponding to a fault prediction result, to generate a dataset of real-time data and prediction results.

[0186] Step S162: Extract samples from the real-time data and prediction result datasets that match the actual fault conditions of intelligent driving component fault prediction results, filter out the real-time operating status data of intelligent driving multi-components with accurate predictions, and generate an effective prediction sample set.

[0187] The fault prediction results in the real-time data and prediction result dataset are compared with the actual fault situations. For samples where the prediction results match the actual fault situations, the predictions are considered accurate. The corresponding real-time operating status data of multiple intelligent driving components are then filtered out to form an effective prediction sample set. These samples can reflect the operating status characteristics of the components at the time of accurate prediction.

[0188] Step S163: Extract the operating status features and multi-component coupling relationship information corresponding to the real-time operating status data of multiple components under different intelligent driving modes and different environmental conditions from the effective prediction sample set, and generate effective operating status feature data.

[0189] The effective prediction sample set is analyzed and classified according to different intelligent driving modes and environmental conditions. Under each category, operational status features corresponding to the real-time operating status data of multiple components are extracted, such as the operating temperature range and vibration frequency characteristics of each component, as well as the coupling relationship information between multiple components, such as coupling strength and direction of influence. The extracted features and information are integrated to generate effective operational status feature data.

[0190] Step S164: Extract the adjusted dynamic coupling model, multi-component coupling relationship data, and component operation status data classified by mode from the multi-component dynamic coupling modeling data in the intelligent driving scenario.

[0191] From the existing dynamic coupling modeling data of multiple components in intelligent driving scenarios, we extract the adjusted dynamic coupling model, which is the model previously constructed to describe the dynamic coupling law of components; multi-component coupling relationship data, which includes coupling relationship information between components; and component operating status data classified by mode, that is, component operating status data under different intelligent driving modes.

[0192] Step S165: Compare the effective operating status feature data with the multi-component coupling relationship data, output the coupling relationship features not included in the multi-component coupling relationship data, and generate new coupling relationship data.

[0193] The multi-component coupling relationship information in the effective operating status feature data is compared with the multi-component coupling relationship data. Coupling relationship features present in the effective operating status feature data but not in the multi-component coupling relationship data are identified; these are new, previously undiscovered coupling relationships. These newly added coupling relationship features are extracted to generate new coupling relationship data.

[0194] Step S166: Analyze the coupling influence strength, direction, and carrier between each component in the newly added coupling relationship data, determine the type and transmission law of the newly added coupling relationship, and generate the newly added coupling influence parameter data.

[0195] Each coupling relationship in the newly added coupling relationship data is analyzed to determine the coupling influence strength (i.e., the magnitude of the influence) between each component; the direction of influence (whether it is unidirectional or bidirectional); and the influencing carrier (whether it is a mechanical structure, hydraulic pipeline, or other carrier). Based on this information, the type of the newly added coupling relationship is determined, such as mechanical coupling or electrical coupling, and its transmission rules are summarized, such as transmission speed and transmission conditions. These parameters are then compiled to generate the newly added coupling influence parameter data.

[0196] Step S167: Add the newly added coupling relationship data and the newly added coupling influence parameter data to the multi-component coupling relationship data, update the content of the multi-component coupling relationship data, and generate the updated multi-component coupling relationship data.

[0197] The newly added coupling relationship data and newly added coupling influence parameter data are integrated into the original multi-component coupling relationship data, the missing coupling relationship information is supplemented, the content of the multi-component coupling relationship data is updated, and the updated multi-component coupling relationship data is generated.

[0198] Step S168: Adjust the parameters of the adjusted dynamic coupling model based on the updated multi-component coupling relationship data to generate the updated dynamic coupling model.

[0199] The parameters of the adjusted dynamic coupling model are set based on the original multi-component coupling relationship data. Now, the model parameters are adjusted according to the updated multi-component coupling relationship data. For example, for newly added coupling relationships, corresponding connection weights and influence coefficients are added to the model; for changes in the influence intensity of existing coupling relationships, the corresponding parameter values ​​in the model are adjusted. Through these adjustments, an updated dynamic coupling model is generated, enabling it to more accurately describe the dynamic coupling laws between components.

[0200] Step S169: Collect historical fault data corresponding to the newly added intelligent driving core components, associate and bind the newly added historical fault data with the updated multi-component coupling relationship data and the updated dynamic coupling model, and generate updated preliminary coupling modeling data.

[0201] Over time, new historical fault data for core components of intelligent driving will be generated. This newly added historical fault data is collected and associated with updated multi-component coupling relationship data and updated dynamic coupling models. This allows the new fault data to be combined with existing coupling relationships and models to generate updated preliminary coupling modeling data.

[0202] Step S1610: Integrate the updated preliminary coupling modeling data with the effective operating state feature data to generate updated multi-component dynamic coupling modeling data under the intelligent driving scenario. Input the updated multi-component dynamic coupling modeling data under the intelligent driving scenario into the artificial intelligence model to replace the original modeling data, update the model's learning parameters and coupling relationship recognition logic, and realize the dynamic adaptation of the artificial intelligence model to the intelligent driving working conditions.

[0203] The updated preliminary coupling modeling data is integrated with the effective operational state feature data, merging and unifying their content to generate updated multi-component dynamic coupling modeling data for intelligent driving scenarios. This updated modeling data is then input into the artificial intelligence model, replacing the original modeling data. Upon receiving the new data, the model relearns and retrains, updating its internal learning parameters and coupling relationship recognition logic to better adapt to current intelligent driving conditions and achieve dynamic adaptation. Through this iterative update process, the fault prediction accuracy and adaptability of the artificial intelligence model are continuously improved.

[0204] In one exemplary embodiment, an AI-based automotive component fault prediction system is provided. This AI-based automotive component fault prediction system can be a terminal, server, etc., and its internal structure diagram can be as follows: Figure 2As shown, this AI-based automotive component fault prediction system includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides the environment for the operation of the operating system and computer programs in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, near-field communication, or other technologies. When the computer program is executed by the processor, it implements an AI-based automotive component fault prediction method. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, or a button, trackball, or touchpad set on the housing of an AI-based automotive parts fault prediction system, or an external keyboard, touchpad, or mouse, etc.

[0205] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A method for predicting automotive component failures based on artificial intelligence, characterized in that, The method includes: The environmental perception module and component status monitoring module of the intelligent driving system are linked to construct dynamic coupling modeling data of multiple components in intelligent driving scenarios. Based on the dynamic coupling modeling data of multiple components in the intelligent driving scenario, the chain transmission logic of faults among the core components of intelligent driving is analyzed, and the chain transmission path data of faults in intelligent driving components is generated. By integrating real-time environmental data of intelligent driving with real-time operating status data of multiple components, and dynamically associating it with fault chain transmission path data of intelligent driving components, real-time coupled status data of multiple intelligent driving components is generated. Based on the real-time coupled status data of multiple intelligent driving components and the fault chain transmission path data of intelligent driving components, the evolution process of faults under intelligent driving conditions is deduced, and the fault prediction results of intelligent driving components are generated. By leveraging the correspondence between real-time operating status data of multiple intelligent driving components and fault prediction results of intelligent driving components, the dynamic coupling modeling data of multiple components in intelligent driving scenarios is iteratively updated to achieve dynamic adaptation of artificial intelligence models to intelligent driving conditions. The method, based on real-time coupled state data of multiple intelligent driving components and fault chain propagation path data of intelligent driving components, deduces the evolution process of faults under intelligent driving conditions and generates fault prediction results for intelligent driving components, including: Real-time dynamic coupling data, path position matching data, state evolution trend data, control mode information, power output state information, and path planning information are extracted from the real-time coupling state data of multiple components in intelligent driving. Extract collaborative adaptation fault chain transmission path, fault chain transmission feature data, and fault key transmission node data from the fault chain transmission path data of intelligent driving components. Based on path location matching data, determine the fault chain transmission stage corresponding to the current multi-component coupled operation state, and generate data for the current transmission stage. By combining state evolution trend data and fault chain transmission characteristic data, the fault transmission process in the current transmission stage is deduced, the next component to which the fault will be transmitted, the time required for transmission, and the changes in control signals during the transmission process are predicted, and fault transmission prediction data is generated. Based on real-time dynamic coupling data and control mode information, the impact of the coupling effect of each component on the fault evolution during the fault transmission process is analyzed. Combined with the path planning information of intelligent driving, the impact trend of fault evolution on driving safety is judged, and fault evolution trend judgment data is generated. Based on the data of critical fault transmission nodes, the arrival time, corresponding status and multi-component collaborative anomalies of each critical node in the subsequent transmission process of the fault are predicted, and critical node prediction data is generated. By combining fault propagation prediction data, fault evolution trend judgment data, and key node prediction data, the final range of components that may be involved in the fault, the final time of the fault occurrence, and the impact of the fault on the switching of intelligent driving modes are determined, and the final state prediction data of the fault is generated. Extract the core prediction information from the final state prediction data of the fault, including the list of parts involved in the fault, the time of the fault occurrence, the division of the fault evolution stages, and the scope of intelligent driving functions affected by the fault, and generate core prediction data. The core prediction data is associated with the corresponding real-time environmental data of intelligent driving and the real-time operating status data of multiple components. The background of intelligent driving conditions corresponding to the prediction results is labeled, and associated labeled prediction data is generated. The associated labeled prediction data is standardized and organized to generate intelligent driving component fault prediction results that include fault type, involved component links, occurrence time, evolution process, and scope of affected functions.

2. The method for predicting automotive component failures based on artificial intelligence according to claim 1, characterized in that, The environmental perception module and component status monitoring module of the linked intelligent driving system construct dynamic coupling modeling data of multiple components in the intelligent driving scenario, including: The environmental perception module of the intelligent driving system collects environmental data during the intelligent driving process. The environmental data during the intelligent driving process includes road type data, traffic flow data, meteorological condition data, lighting condition data, and obstacle distribution data. The component status monitoring module of the intelligent driving system collects the operating status data of the core components of intelligent driving. The operating status data of the core components of intelligent driving includes steering system component data, braking system component data, power system component data, perception system component data, and decision control system component data. Each system component data includes operating temperature data, operating pressure data, vibration data, energy consumption data, running time data, and control signal interaction data. Collect historical fault data corresponding to core components of intelligent driving. The historical fault data corresponding to core components of intelligent driving includes fault type data, intelligent driving mode data when the fault occurred, environmental data when the fault occurred, component link data involved in the fault, and control response change data induced by the fault. The operating status data of core components of intelligent driving are classified according to intelligent driving modes, including highway automatic driving mode, urban road automatic driving mode, traffic jam following mode, curve driving mode, and parking mode, and component operating status data are generated according to the mode classification. The system correlates environmental data during intelligent driving with component operating status data categorized by mode, mines the correlation of changes in the operating status of each component under different environmental conditions, and generates environment-component correlation data. Extract the sequence of changes in the operating status of components before the occurrence of a fault from the historical fault data corresponding to the core components of intelligent driving, and generate the fault pre-fault state change sequence data. By fusing component operating status data, environment-component correlation data, and fault pre-state change sequence data classified by mode, the coupling relationship of multiple component operating status under intelligent driving mode is explored to generate multi-component coupling relationship data. The system inputs the component operation status data, environment-component correlation data, fault pre-fault state change sequence data, and multi-component coupling relationship data into the artificial intelligence modeling model. The artificial intelligence modeling model learns the dynamic coupling law of multiple components under different intelligent driving modes and different environmental conditions, and generates an initial dynamic coupling model. The parameters of the initial dynamic coupling model are adjusted according to the control logic of the intelligent driving system, so that the adjusted dynamic coupling model can adapt to the interaction characteristics of components and control commands in the intelligent driving process. The adjusted dynamic coupling model is then integrated with all related data to generate multi-component dynamic coupling modeling data in the intelligent driving scenario.

3. The method for predicting automotive component failures based on artificial intelligence according to claim 2, characterized in that, The data, based on the dynamic coupling modeling of multiple components in the intelligent driving scenario, analyzes the chain transmission logic of faults among core intelligent driving components, and generates fault chain transmission path data for intelligent driving components, including: Extract the adjusted dynamic coupling model, multi-component coupling relationship data, and fault pre-state change sequence data from the multi-component dynamic coupling modeling data in intelligent driving scenarios; Based on the adjusted dynamic coupling model, the transmission triggering conditions of different fault types among the core components of intelligent driving are analyzed, and fault transmission triggering condition data is generated. By combining multi-component coupling relationship data, connection link information of core intelligent driving components is extracted. The connection links of core intelligent driving components include mechanical transmission links, hydraulic transmission links, electrical control links, and signal interaction links, thereby generating component connection link data. Extract the state features of the fault-initiating component from the fault pre-fault state change sequence data, determine the component where the fault first occurred and the initial fault state, and generate fault initiation source data; Based on component connection link data and fault transmission triggering condition data, the order in which the fault is transmitted from the originating component to other related components is deduced based on preset logical rules. The transmission order information of the fault in different links is extracted to generate fault link transmission order data. Track the state changes of faults during the transmission process between components, record the changes in state parameters and control signal responses of each component during the fault transmission process, and generate fault transmission state change data. Mark key transmission nodes in the fault transmission process and generate fault key transmission node data. The key transmission nodes include fault cross-link transmission nodes, fault aggravation nodes, and fault-induced multi-component collaborative anomaly nodes. By concatenating the fault origin data, fault link transmission sequence data, and fault key transmission node data in the order of fault transmission time, a preliminary fault chain transmission path is constructed. By combining the power transmission logic and control signal interaction logic of the intelligent driving system, the initial fault chain transmission path is adjusted so that the adjusted fault chain transmission path conforms to the actual collaborative working logic of the components in the intelligent driving process, and a collaboratively adapted fault chain transmission path is generated. Extract the transmission feature information in the collaborative adaptation fault chain transmission path. The transmission feature information includes the transmission duration of the fault in each link, the state change rate during the fault transmission process, and the control signal delay time during fault transmission, and generate fault chain transmission feature data. By associating and integrating the collaborative adaptation fault chain transmission path with the fault chain transmission characteristic data, fault chain transmission path data for intelligent driving components is generated.

4. The method for predicting automotive component failures based on artificial intelligence according to claim 3, characterized in that, The system integrates real-time environmental data of intelligent driving with real-time operational status data of multiple components, and dynamically correlates this data with fault chain transmission path data of intelligent driving components to generate real-time coupled status data of multiple intelligent driving components, including: The intelligent driving system collects real-time environmental data through its environmental perception module. This real-time environmental data includes real-time road type data, real-time traffic flow data, real-time weather condition data, real-time lighting condition data, and real-time obstacle distribution data. The intelligent driving system collects real-time operating status data of multiple components through the component status monitoring module. The real-time operating status data of multiple components includes real-time component data of the steering system, braking system, power system, perception system, and decision control system. The real-time component data of each system includes real-time operating temperature data, real-time operating pressure data, real-time vibration data, real-time energy consumption data, real-time running time data, and real-time control signal interaction data. The real-time environmental data of intelligent driving and the real-time operating status data of multiple components are input into the real-time data fusion module of the artificial intelligence model for data time-series alignment processing, so that the environmental data and the operating status data of components correspond precisely in the time dimension, and generate time-series aligned real-time fused data. Real-time operating status features of each core component of intelligent driving are extracted from time-aligned real-time fused data to generate real-time status feature data of a single component. Extract collaborative adaptation fault chain transmission paths and fault chain transmission feature data from the fault chain transmission path data of intelligent driving components. Analyze the correlation between the real-time status feature data of each individual component, mine the dynamic coupling information between each component under the current intelligent driving condition, and generate real-time dynamic coupling data. The real-time status feature data of a single component is compared with the fault transmission status change data of the corresponding component in the collaborative adaptation fault chain transmission path to locate the corresponding position of the current operating status of each component in the fault chain transmission path and generate path position matching data. By combining the transmission duration, state change rate, and control signal delay time in the fault chain transmission characteristic data, we can analyze the trend of the current operating state of each component evolving into a fault state and generate state evolution trend data. By integrating real-time dynamic coupling data, path location matching data, and state evolution trend data, preliminary real-time coupling state data is formed. The system supplements the current control mode information, power output status information, and path planning information of the intelligent driving system into the preliminary real-time coupling status data, generates real-time coupling status data after control supplementation, and performs structured integration of the real-time coupling status data after control supplementation to generate real-time coupling status data of multiple intelligent driving components.

5. The method for predicting automotive component failures based on artificial intelligence according to claim 3, characterized in that, The extraction of connection link information of core intelligent driving components by combining multi-component coupling relationship data includes: Extract information on the coupling influence relationships, influence methods, and influence carriers among the core components of intelligent driving from multi-component coupling relationship data; Based on the information of the carrier, different types of connection links are distinguished. The coupling influence relationship with mechanical structure as carrier is classified as mechanical transmission link related data, and the coupling influence relationship with hydraulic pipeline as carrier is classified as hydraulic transmission link related data. The coupling effects carried by electrical lines are classified as electrical control link related data, and the coupling effects carried by data signals are classified as signal interaction link related data. The data related to the mechanical transmission link is refined by extracting information on component composition, transmission ratio, and connection points to form the composition and connection details of the mechanical transmission link, thus generating detailed data on the mechanical transmission link. The data related to the hydraulic transmission link is refined by extracting information on the composition of hydraulic components, the flow path of hydraulic oil, and the pressure transmission nodes to form the composition and flow details of the hydraulic transmission link and generate detailed data of the hydraulic transmission link. The data related to the electrical control link is refined by extracting information on the composition of electrical components, current transmission paths, and control signal triggering nodes to form the composition and transmission details of the electrical control link and generate detailed data of the electrical control link. The data related to the signal interaction link is refined by extracting signal type information, signal transmission protocol information, and signal receiving and feedback node information to form the composition and interaction details of the signal interaction link and generate detailed data of the signal interaction link. By integrating detailed data on mechanical transmission links, hydraulic transmission links, electrical control links, and signal interaction links, a complete connection link network for core components of intelligent driving is constructed. Label the intersection nodes of each link in the link network, that is, the components or connection parts shared by different types of links, and generate link intersection node data; By associating the complete connection link network with the link intersection node data, component connection link data containing detailed composition, connection relationship and intersection node of each link is generated.

6. The method for predicting automotive component failures based on artificial intelligence according to claim 4, characterized in that, The analysis examines the correlation between the real-time state characteristic data of each individual component, mines the dynamic coupling information between components under the current intelligent driving condition, and generates real-time dynamic coupling data, including: Extract real-time operating temperature features, real-time operating pressure features, real-time vibration features, real-time energy consumption features, and real-time control signal interaction features from the real-time status feature data of each individual component to generate a multi-dimensional real-time status feature set for each component. The core components of the intelligent driving decision control system are selected as benchmark components, and the multi-dimensional real-time state feature set of the benchmark components is extracted to generate a benchmark feature set. The multi-dimensional real-time state feature sets of other core intelligent driving components are compared one by one with the benchmark feature set. The synchronicity of changes in the same dimension features between different components is analyzed, with a focus on the synchronous response of control signal interaction features, and feature synchronization change data is generated. Calculate the degree of mutual influence between the reference component and other components in the same dimension, determine the strength of the effect of a change in a certain dimension of the reference component on the same dimension of other components, and generate feature influence strength data. Replace the benchmark component with the core component of the intelligent driving power system, and repeat the above process of benchmark feature selection, feature comparison, synchronization analysis, and influence intensity calculation to generate a second set of feature synchronization change data and feature influence intensity data. The benchmark component is replaced again with the core component of the intelligent driving steering system. The above process is repeated to generate a third set of feature synchronous change data and feature influence intensity data. By integrating three sets of feature synchronization change data and feature influence intensity data, a multi-dimensional feature coupling relationship network between various core components of intelligent driving is constructed, feature coupling network data is generated, and coupling relationships that meet the conditions are selected from the feature coupling network data according to the preset coupling strength threshold and synchronization threshold, generating active coupling relationship data. Analyze the direction of each coupling relationship in the active coupling relationship data to determine whether it is a unidirectional influence or a bidirectional mutual influence, extract information on the initiating component and the affected component, and generate coupling direction data; The active coupling relationship data is matched with the pre-stored coupling relationship mechanism knowledge base to generate coupling mechanism explanation data; By associating and integrating active coupling relationship data, coupling direction data, and coupling mechanism description data, real-time dynamic coupling data is generated.

7. The method for predicting automotive component failures based on artificial intelligence according to claim 4, characterized in that, The method combines state evolution trend data and fault chain propagation characteristic data to deduce the subsequent fault propagation process in the current propagation stage, predict the next component to which the fault will be propagated, the time required for propagation, and the changes in control signals during the propagation process, generating fault propagation prediction data, including: Extract the rate of change, direction of change, and trend of change of control signal response of each component from the current state to the fault state from the state evolution trend data, and generate state evolution parameter data. Extract the fault propagation duration, state change rate, control signal delay time, and fault propagation direction from the fault chain propagation feature data to generate fault propagation parameter data. Determine the component where the current fault is located, the current fault status parameters, the current fault transmission progress, and the current link type from the current transmission stage data, and generate the current fault status data; Based on the current fault status data and fault transmission parameter data, determine the transmission conditions for the fault to be transmitted from the current component to the next related component, extract the threshold values ​​of the status parameters and the threshold values ​​of the control signal changes required for transmission, and generate fault transmission condition data. Analyze the rate of change of the current component's state and the trend of change of the control signal response in the state evolution parameter data, determine whether the current fault state of the component meets the fault propagation conditions, and generate data on the satisfaction of the propagation conditions. If the transmission conditions are met, the transmission time required for fault transmission is calculated by combining the transmission duration in the fault transmission parameter data and the coupling influence strength from the current component to the next component. At the same time, the delay time and signal strength change of the control signal during the transmission process are predicted, and transmission time calculation data and control signal change prediction data are generated. Based on the fault transmission direction in the fault transmission parameter data and the correlation in the component connection link data, determine the next component to which the fault will be transmitted, extract the link type information of the next component, and generate the data of the next faulty component. Predict the initial fault state of the next component after receiving the fault transmission, and generate the initial fault state data of the next component by combining the state change rate in the fault transmission parameter data, the current fault state parameters, and the control signal change prediction data. If the transmission conditions are not met, analyze the difference between the current component state evolution parameter data and the fault transmission condition data, predict the time required to meet the transmission conditions, and analyze the possible control signal adjustment trend to meet the transmission conditions, generating transmission condition satisfaction time data and control signal adjustment trend data. The fault transmission prediction data is generated by associating and integrating the data of the next faulty component, the transmission time calculation data, the control signal change prediction data, the initial fault state data of the next component, the transmission condition satisfaction time data, and the control signal adjustment trend data.

8. The method for predicting automotive component failures based on artificial intelligence according to claim 2, characterized in that, The method of iteratively updating the dynamic coupling modeling data of multiple components in intelligent driving scenarios by utilizing the correspondence between real-time operating status data of multiple intelligent driving components and fault prediction results of intelligent driving components, thereby achieving dynamic adaptation of the artificial intelligence model to intelligent driving conditions, includes: Collect real-time operating status data of multiple intelligent driving components and corresponding intelligent driving component fault prediction results within a preset time period to generate a real-time data and prediction result dataset. Extract samples from real-time data and prediction result datasets that match the actual fault conditions of intelligent driving component fault prediction results, filter out accurate real-time operating status data of multiple intelligent driving components, and generate an effective prediction sample set. Extract the operational status features and multi-component coupling relationship information corresponding to the real-time operational status data of multiple components under different intelligent driving modes and environmental conditions from the effective prediction sample set, and generate effective operational status feature data. Extract the adjusted dynamic coupling model, multi-component coupling relationship data, and component operation status data classified by mode from the multi-component dynamic coupling modeling data in intelligent driving scenarios; The effective operating status feature data is compared with the multi-component coupling relationship data, and the coupling relationship features not included in the multi-component coupling relationship data are output to generate new coupling relationship data. Analyze the coupling influence strength, direction, and carrier between components in the newly added coupling relationship data to determine the type and transmission law of the newly added coupling relationship and generate the newly added coupling influence parameter data. Add the new coupling relationship data and the new coupling influence parameter data to the multi-component coupling relationship data, update the content of the multi-component coupling relationship data, and generate the updated multi-component coupling relationship data. The parameters of the adjusted dynamic coupling model are adjusted based on the updated multi-component coupling relationship data to generate the updated dynamic coupling model. Collect historical fault data corresponding to newly added intelligent driving core components, associate and bind the newly added historical fault data with the updated multi-component coupling relationship data and the updated dynamic coupling model, and generate updated preliminary coupling modeling data; The updated preliminary coupling modeling data is integrated with the effective operating state feature data to generate updated multi-component dynamic coupling modeling data under the intelligent driving scenario. The updated multi-component dynamic coupling modeling data under the intelligent driving scenario is then input into the artificial intelligence model to replace the original modeling data, update the model's learning parameters and coupling relationship recognition logic, and realize the dynamic adaptation of the artificial intelligence model to the intelligent driving conditions.

9. An artificial intelligence-based automotive component fault prediction system, characterized in that, include: processor; A machine-readable storage medium for storing machine-executable instructions of the processor; The processor is configured to execute the AI-based automotive component fault prediction method according to any one of claims 1 to 8 by executing the machine-executable instructions.

Images