Multi-agent based vehicle fault diagnosis system, method and apparatus

Abstract

This application provides a vehicle fault diagnosis system, method, and device based on a multi-agent system. The system includes: a data acquisition agent, an event detection agent, a knowledge graph agent, and a fault diagnosis agent. The data acquisition agent collects target vehicle operation data; the event detection agent performs event detection on the target vehicle operation data to obtain event packets; the knowledge graph agent obtains vehicle model information corresponding to the target vehicle and constructs a reference knowledge graph corresponding to the target vehicle based on the vehicle model information; and the fault diagnosis agent performs diagnostic reasoning on the event packets based on the reference knowledge graph to obtain the target diagnostic result. Using the embodiments of this application can improve the real-time performance of vehicle fault diagnosis.

Description

Technical Field

[0001] This application relates to the field of vehicle diagnostic technology, and in particular to a vehicle fault diagnosis system, method and device based on multi-agent technology. Background Technology

[0002] With the popularization of new energy vehicles and intelligent connected vehicles, the functional integration of the vehicle's electronic and electrical systems is becoming increasingly higher, and cross-domain control and centralized electronic and electrical architectures are gradually becoming the mainstream. The number of vehicle sensors, the types of signals, and the frequency of data interaction have increased significantly, resulting in fault modes exhibiting multi-source, chain-like propagation, and dynamic evolution characteristics.

[0003] Currently, traditional diagnostic methods are mostly passive diagnostic approaches that intervene after a fault occurs. They lack an event-driven real-time monitoring and triggering mechanism, making it difficult to capture early signs of faults in the early stages of abnormal changes in the vehicle's critical state. This makes it impossible to achieve early warning and proactive intervention for faults, which can easily lead to the further spread of faults and trigger a chain of problems.

[0004] Therefore, improving the real-time performance of vehicle fault diagnosis is an urgent issue that needs to be addressed. Summary of the Invention

[0005] This application provides a vehicle fault diagnosis system, method, and device based on multi-agent technology, which improves the real-time performance of vehicle fault diagnosis.

[0006] In a first aspect, embodiments of this application provide a vehicle fault diagnosis system based on multiple agents, the system comprising: a data acquisition agent, an event detection agent, a knowledge graph agent, and a fault diagnosis agent, wherein: The data acquisition intelligent agent is used to collect target vehicle operation data of the target vehicle. The event detection agent is used to perform event detection on the target vehicle's operating data to obtain an event packet; The knowledge graph agent is used to obtain vehicle model information corresponding to the target vehicle; and to construct a reference knowledge graph corresponding to the target vehicle based on the vehicle model information. The fault diagnosis agent is used to perform diagnostic reasoning on the event package based on the reference knowledge graph to obtain the target diagnosis result.

[0007] Secondly, embodiments of this application provide a vehicle fault diagnosis method based on multiple agents, the method comprising: Collect target vehicle operation data; Event detection is performed on the target vehicle's operating data to obtain event packets; Obtain the vehicle model information corresponding to the target vehicle; construct a reference knowledge graph corresponding to the target vehicle based on the vehicle model information; The event package is diagnosed and reasoned based on the reference knowledge graph to obtain the target diagnosis result.

[0008] Thirdly, embodiments of this application provide an electronic device, including: a processor and a memory, the memory being used to store one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the processor, the programs including instructions for performing the steps in the second aspect of this application.

[0009] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program for electronic data interchange, wherein the computer program causes a computer to perform some or all of the steps described in the second aspect of this application.

[0010] Fifthly, embodiments of this application provide a computer program product, wherein the computer program product includes a non-transitory computer-readable storage medium storing a computer program operable to cause a computer to perform some or all of the steps described in the second aspect of embodiments of this application. The computer program product may be a software installation package.

[0011] As can be seen, the vehicle fault diagnosis system based on multi-agent provided in this application adopts a distributed architecture of four independent intelligent agents: data acquisition, event detection, knowledge graph, and fault diagnosis. Each intelligent agent collaborates and achieves efficient data flow through standardized interfaces. Among them, the event detection intelligent agent can identify multi-dimensional abnormal events in vehicle operation data, covering scenarios such as threshold exceeding and abnormal time-series patterns. It can capture key precursor signals and generate event packets before a fault occurs, triggering subsequent diagnostic processes, thereby improving the real-time performance of vehicle fault diagnosis. The knowledge graph intelligent agent can dynamically construct a dedicated reference knowledge graph based on vehicle model information, which can quickly adapt to the centralized electronic and electrical architecture of new energy vehicles and intelligent connected vehicles, as well as scenarios such as vehicle configuration changes and component upgrades, significantly reducing system maintenance costs and response cycles. The fault diagnosis intelligent agent performs bidirectional traversal reasoning based on the relational edges of the reference knowledge graph, which can not only output target diagnostic results containing fault type and confidence level, but also generate traceable causal reasoning paths. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application or the background art, the accompanying drawings used in the embodiments of this application or the background art will be described below.

[0013] Figure 1 This is a schematic diagram of the composition of a vehicle fault diagnosis system based on multiple agents provided in an embodiment of this application; Figure 2 This is a schematic diagram of a multi-vehicle communication bus data acquisition method provided in an embodiment of this application; Figure 3 This is a flowchart of an event detection method provided in an embodiment of this application; Figure 4 This is a flowchart of a method for constructing a reference electricity knowledge graph provided in an embodiment of this application; Figure 5 This is a schematic diagram of another vehicle fault diagnosis system based on multiple agents provided in an embodiment of this application; Figure 6 This is an application scenario diagram of a vehicle fault diagnosis system based on multiple agents provided in an embodiment of this application; Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application; Figure 8 This is a flowchart illustrating a vehicle fault diagnosis method based on multiple agents provided in an embodiment of this application. Detailed Implementation

[0014] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present application.

[0015] The terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0016] It should be understood that the term "and / or" in this document is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this document indicates that the preceding and following related objects are in an "or" relationship. In the embodiments of this application, "multiple" refers to two or more.

[0017] In the embodiments of this application, "at least one item" or its similar expression refers to any combination of these items, including any combination of a single item or a plurality of items. "One or more" means one or more, while "multiple" means two or more. For example, "at least one item" of a, b, or c can represent the following seven cases: a, b, c; a and b; a and c; b and c; a, b, and c. Each of a, b, and c can be an element or a set containing one or more elements.

[0018] In this application, the term "connection" refers to various connection methods, such as direct connection or indirect connection, to achieve communication between devices. This application does not impose any limitations on this.

[0019] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0020] With the popularization of new energy vehicles and intelligent connected vehicles, the functional integration of the vehicle's electronic and electrical systems is becoming increasingly higher, and cross-domain control and centralized electronic and electrical architectures are gradually becoming the mainstream. The number of vehicle sensors, the types of signals, and the frequency of data interaction have increased significantly, resulting in fault modes exhibiting multi-source, chain-like propagation, and dynamic evolution characteristics.

[0021] Currently, traditional diagnostic methods are mostly passive diagnostic approaches that intervene after a fault occurs. They lack an event-driven real-time monitoring and triggering mechanism, making it difficult to capture early signs of faults in the early stages of abnormal changes in the vehicle's critical state. This makes it impossible to achieve early warning and proactive intervention for faults, which can easily lead to the further spread of faults and trigger a chain of problems.

[0022] Therefore, improving the real-time performance of vehicle fault diagnosis is an urgent issue that needs to be addressed.

[0023] To address the aforementioned problems, this application provides a vehicle fault diagnosis system, method, and device based on a multi-agent framework. The multi-agent vehicle fault diagnosis system includes: a data acquisition agent, an event detection agent, a knowledge graph agent, and a fault diagnosis agent. Specifically: the data acquisition agent collects target vehicle operating data; the event detection agent performs event detection on the target vehicle operating data to obtain event packets; the knowledge graph agent obtains vehicle model information corresponding to the target vehicle and constructs a reference knowledge graph corresponding to the target vehicle based on the vehicle model information; and the fault diagnosis agent performs diagnostic reasoning on the event packets based on the reference knowledge graph to obtain a target diagnostic result. Therefore, by using this system, the real-time performance of vehicle fault diagnosis is improved.

[0024] For easier understanding, please refer to Figure 1 , Figure 1 This is a schematic diagram illustrating the composition of a multi-agent-based vehicle fault diagnosis system provided in an embodiment of this application. The system includes: a data acquisition agent, an event detection agent, a knowledge graph agent, and a fault diagnosis agent, wherein: The data acquisition agent is used to collect target vehicle operation data.

[0025] Optionally, regarding the collection of target vehicle operation data, the data collection agent is specifically used for: A1. Obtain multiple vehicle communication buses corresponding to the target vehicle; A2. Data is collected from the target vehicle through the multiple vehicle communication buses to obtain multiple initial vehicle operation data; each vehicle communication bus corresponds to one initial vehicle operation data. A3. Perform preprocessing operations on each of the multiple initial vehicle operation data to obtain multiple reference vehicle operation data; A4. Encapsulate the multiple reference vehicle operation data according to a preset signal flow structure to obtain the target vehicle operation data; the target vehicle operation data includes multiple signal flows, each signal flow corresponding to one reference vehicle operation data.

[0026] In this embodiment, the vehicle communication bus includes at least two of the following: CAN bus, Ethernet bus, and LIN bus, without specific limitations. The CAN bus is used to transmit critical real-time data such as power control and chassis control (e.g., engine speed, brake pressure); the Ethernet bus is used to transmit large amounts of non-real-time data (e.g., in-vehicle entertainment system data, camera image data); and the LIN bus is used to transmit low-speed data such as vehicle body electronics (e.g., window control, lighting status data). The target vehicle can be a new energy vehicle or an intelligent connected vehicle, without specific limitations.

[0027] In a specific embodiment, the target vehicle includes multiple system modules (such as a powertrain control system, chassis control system, and body electronic system). Different system modules typically use different types of on-board communication buses for data transmission. The data acquisition agent first establishes communication with the bus management unit of the target vehicle through a vehicle-side diagnostic interface (such as an OBD interface) to obtain the bus configuration information of the target vehicle, and determines multiple on-board communication buses and communication parameters based on the bus configuration information.

[0028] Next, the data acquisition agent, based on communication parameters (such as baud rate and communication protocol type), employs a parallel acquisition mechanism to synchronously acquire data from different system modules of the target vehicle through multiple onboard communication buses, obtaining multiple initial vehicle operation data. Each onboard communication bus corresponds to acquiring initial vehicle operation data for one type of system module. These multiple initial vehicle operation data include, but are not limited to: diagnostic fault codes, freeze frames, real-time sensor streams (such as engine speed, coolant temperature, battery voltage, motor current, brake pressure, etc.), control command status, and actuator feedback signals; specific limitations are not specified here.

[0029] Then, preprocessing operations are performed on each of the multiple initial vehicle operation data sets to obtain multiple reference vehicle operation data sets. These preprocessing operations include, but are not limited to, time synchronization, outlier removal, dimensional normalization, and data compression. Specifically, a pre-defined Kalman filter algorithm can be used to smooth the initial vehicle operation data, filtering out high-frequency noise from electromagnetic interference. For example, Kalman filtering of engine speed data yields a smooth speed time-series curve. Outliers in the initial vehicle operation data can also be identified and removed. For instance, values ​​where battery voltage jumps outside the normal range (e.g., 2.5V-4.2V) can be removed, and linear interpolation of data from preceding and following times can be used to complete the data. Furthermore, the clock signal of the target vehicle can be used as a reference to timestamp and calibrate the initial vehicle operation data collected from different buses, ensuring consistency in the time dimension of all data.

[0030] Finally, the data acquisition agent encapsulates the operating data of multiple reference vehicles according to a pre-defined standardized signal flow structure to obtain the operating data of the target vehicle. The target vehicle operating data includes multiple signal flows, each corresponding to one set of reference vehicle operating data. This signal flow structure includes, but is not limited to: signal ID, acquisition timestamp, numerical sequence, data type, and data source; specific limitations are not specified here.

[0031] It is evident that through the design of multi-bus parallel acquisition, standardized preprocessing, and structured encapsulation, the data acquisition agent can efficiently acquire multi-source heterogeneous operating data of the target vehicle, and the processed data possesses real-time performance, completeness, and uniformity.

[0032] For easier understanding, please refer to Figure 2 , Figure 2 This is a schematic diagram of multi-vehicle communication bus data acquisition provided in an embodiment of this application. The target vehicle includes a power control system, an autonomous driving system, and a body electronic system. The corresponding system modules can be acquired through the first vehicle communication bus, the second vehicle communication bus, and the third vehicle communication bus to obtain first initial vehicle operation data, second initial vehicle operation data, and third initial vehicle operation data. The first initial vehicle operation data, the second initial vehicle operation data, and the third initial vehicle operation data are then aggregated into a data acquisition agent to complete the data acquisition.

[0033] The event detection agent is used to perform event detection on the target vehicle's operating data to obtain event packets.

[0034] Optional, please refer to Figure 3 , Figure 3 This is a flowchart of an event detection method provided in an embodiment of this application. In the step of performing event detection on the target vehicle's operating data to obtain an event packet, the event detection agent is specifically used to execute... Figure 3 The steps shown are as follows: B1. Determine the abnormal pattern feature library corresponding to the target vehicle's operating data; B2. Determine multiple signal threshold ranges corresponding to the multiple signal streams in the target vehicle's operating data; each signal stream corresponds to one signal threshold range. B3. If the signal value of at least one first signal stream in the plurality of signal streams exceeds its corresponding signal threshold range, then at least one threshold event is generated based on the at least one first signal stream. B4. If at least one of the multiple signal streams has a signal feature that matches the abnormal pattern feature library, then at least one pattern event is generated based on the at least one second signal stream. B5. Determine the event package based on the at least one threshold event and / or the at least one pattern event.

[0035] In this embodiment of the application, the abnormal pattern feature library is a structured feature set classified by vehicle type. The abnormal pattern feature library stores typical abnormal time sequence feature templates corresponding to each signal stream. Each template includes feature type (such as fluctuation amplitude, peak distribution, and change trend), feature parameters (such as fluctuation frequency range, peak threshold, and trend slope), and corresponding fault association label.

[0036] In a specific embodiment, firstly, the event detection agent retrieves a preset abnormal pattern feature library based on the target vehicle's model information. Then, based on a preset signal configuration table, it matches the signal ID of each signal stream in multiple signal streams to obtain multiple signal threshold ranges. These signal threshold ranges represent the signal safety boundaries under normal vehicle operation, and are divided into upper and lower thresholds. These thresholds are preset by the vehicle manufacturer based on component performance parameters and safe operating specifications, and support automated dynamic updates.

[0037] Next, if the signal value corresponding to any timestamp of a certain signal stream (denoted as the first signal stream) exceeds its bound signal threshold range (e.g., the coolant temperature signal value at the first moment is 96.2℃, exceeding the threshold range of 80℃-95℃), then the first signal stream is determined to have a threshold anomaly. Further, key information corresponding to the first signal stream is extracted, including the signal ID of the first signal stream, the abnormal signal value, the timestamp of the anomaly occurrence, and the threshold range boundary value, and a threshold event is generated based on this key information. Simultaneously, it is determined whether the signal characteristics of at least one second signal stream among the multiple signal streams match the anomaly pattern feature library. If so, at least one pattern event is generated based on the at least one second signal stream; otherwise, no processing is performed.

[0038] Finally, at least one threshold event and / or at least one pattern event are integrated to obtain an event package. Specifically, if only threshold events exist, all threshold events are integrated into an event package; if only pattern events exist, all pattern events are integrated into an event package; if both threshold events and pattern events exist, all events are sorted according to their occurrence timestamps and then integrated into an event package.

[0039] It is evident that by performing event detection on signal stream data, it is possible to quickly identify explicit anomalies (i.e., threshold events) where signal values ​​exceed the standard, and to accurately capture latent fault precursors (i.e., pattern events) where signal timing patterns are abnormal, thus overcoming the limitations of traditional single threshold detection. At the same time, by matching vehicle model information with a dedicated abnormal pattern feature library and signal threshold range, the vehicle model adaptability of event detection is improved.

[0040] Optionally, regarding the statement that if at least one second signal stream among the plurality of signal streams matches the abnormal pattern feature library, the event detection agent is specifically used for: C1. According to the preset sliding window mechanism, extract continuous time segments of each of the multiple signal streams to obtain multiple continuous time segments; C2. Extract the features of multiple reference signals corresponding to the multiple time segments; C3. Obtain the multiple abnormal signal features corresponding to the multiple reference signal features in the abnormal pattern feature library; C4. Calculate the matching degree between the multiple reference signal features and the multiple abnormal signal features according to the preset dynamic time warping algorithm, and obtain multiple matching degrees; C5. Obtain the matching degree that is greater than the preset matching degree threshold among the multiple matching degrees, and obtain at least one matching degree; C6. Determine at least one signal stream corresponding to at least one matching degree among the plurality of signal streams as at least one second signal stream.

[0041] In this embodiment, the core parameters of the sliding window can be dynamically configured according to the signal type. These core parameters include, but are not limited to, window size, sliding step size, and window type. No specific limitations are made here.

[0042] In a specific embodiment, the event detection agent first employs a preset sliding window mechanism to extract continuous time segments from each signal stream in the target vehicle's operational data, resulting in multiple continuous time segments. Then, it extracts reference signal features corresponding to each of these continuous time segments, obtaining multiple reference signal features. These reference signal features are core features capable of characterizing the temporal variation patterns of the signals, such as basic features (signal mean, variance, maximum, and minimum values ​​for each time segment), trend features, and temporal variation features. Finally, based on the signal ID corresponding to each continuous time segment, it retrieves all abnormal signal features (i.e., abnormal signal feature templates) corresponding to that signal ID from an abnormal pattern feature library, thereby obtaining multiple abnormal signal features.

[0043] Next, the matching degree between multiple reference signal features and multiple anomalous signal features is calculated according to a preset dynamic time warping algorithm, resulting in multiple matching degrees. Then, these matching degrees are filtered, retaining those with values ​​greater than a preset matching degree threshold, thus obtaining at least one matching degree. At least one signal stream corresponding to at least one matching degree among the multiple signal streams is marked as a second signal stream, resulting in at least one second signal stream.

[0044] It is evident that by using a sliding window to ensure the integrity of temporal features and multi-dimensional feature extraction to ensure the comprehensiveness of anomaly representation, accurate identification of abnormal patterns is achieved. Simultaneously, by accurately matching anomalous signal features using signal IDs, the reliability of pattern detection is improved.

[0045] In one possible embodiment, the event detection agent can perform collaborative analysis of the changing trends of different signal streams based on preset multi-signal association logic rules. Specifically, if different signals exhibit synchronous changes that do not conform to the normal operating logic of the vehicle, such as "after the vehicle controller issues a 'water pump start' actuator action command, the corresponding water pump speed sensor does not produce the expected speed increase response," or "after the brake pedal travel signal is triggered, the brake pressure sensor value does not increase synchronously," then it is determined that there is an abnormal correlation between the multiple signals, constituting a correlated event. Algorithms such as Pearson correlation coefficient and Granger causality test can be used to quantify the correlation and causality between signals. When the correlation between signals is below a normal threshold range, a correlated event is automatically generated, and key information such as the timestamp of the signal coordination anomaly and the type of logical deviation is recorded.

[0046] The knowledge graph agent is used to obtain vehicle model information corresponding to the target vehicle; and to construct a reference knowledge graph corresponding to the target vehicle based on the vehicle model information.

[0047] Optional, please refer to Figure 4 , Figure 4 This is a flowchart illustrating a method for constructing a reference electricity consumption knowledge graph according to an embodiment of this application. In the step of constructing the reference knowledge graph corresponding to the target vehicle based on the vehicle model information, the knowledge graph agent is specifically used to execute... Figure 4 The steps shown are as follows: D1. Determine multiple nodes corresponding to the target vehicle based on the vehicle model information; the node types of the multiple nodes include: component nodes, signal nodes, and fault mode nodes; D2. Obtain the fault case information corresponding to the vehicle model information; D3. Determine the edge relationship between any two nodes among the multiple nodes based on the fault case information to obtain the edge set; D4. Construct the reference knowledge graph based on the multiple nodes and the set of edges.

[0048] In this embodiment of the application, the corresponding vehicle model information can be synchronously obtained through the vehicle-side diagnostic interface of the target vehicle or the cloud vehicle database. The vehicle model information includes, but is not limited to: vehicle hardware configuration list, electronic and electrical architecture information, sensor parameter table, and control logic description, which are not specifically limited here.

[0049] In a specific embodiment, the knowledge graph agent first performs structured parsing of the vehicle model information to determine multiple nodes corresponding to the target vehicle. These multiple nodes include: component nodes, signal nodes, and fault mode nodes. Component nodes represent the physical hardware components and control modules of the target vehicle, such as the "engine ECU," "water pump," and "battery management system" of the powertrain. Signal nodes represent the sensor-collected signals or control command signals corresponding to the component nodes, such as "water pump speed signal," "coolant temperature signal," and "brake pressure signal." Fault mode nodes represent the possible fault types of each component node, such as "water pump jamming," "engine ECU communication failure," "coolant leakage," and "brake pressure sensor drift," without specific limitations.

[0050] Next, fault case information matching the target vehicle's model information is retrieved. This fault case information consists of actual repair records for that model, including fault symptoms (e.g., "coolant temperature too high alarm"), fault causes (e.g., "water pump control module failure"), related components (e.g., "water pump," "power module"), related signal anomalies (e.g., "water pump speed signal is 0"), and repair plans. Then, semantic parsing and association mining are performed on the structured fault case information, and the edge relationships between any two nodes are determined. These edge relationships include causal and dependency relationships. Finally, a reference knowledge graph is constructed based on the set of multiple nodes and edges.

[0051] It should be noted that causal edges are used to represent the logical relationship that "a change in the state of node A directly leads to a change in the state of node B." For example, from the case of "water pump control module failure, water pump speed signal of 0, coolant temperature signal increase," a causal edge can be extracted stating that "water pump control module (component node) points to water pump speed signal (signal node)" (i.e., a failure of the water pump control module causes an abnormal water pump speed signal). Dependency edges, on the other hand, are used to represent the logical relationship that "the normal operation of node A depends on the functional support of node B." For example, from the case of "power module power supply abnormality, water pump control module cannot work," a dependency edge can be extracted stating that "water pump control module (component node) points to power module (component node)" (i.e., the water pump control module depends on the power module for power supply).

[0052] It is evident that precise node selection through vehicle model information parsing avoids redundancy and bias in general knowledge graphs; edge relationships mined from fault case information ensure the authenticity and reliability of knowledge. Furthermore, the constructed reference knowledge graph provides accurate knowledge support for the subsequent traversal and reasoning of the fault diagnosis agent, and possesses dynamic update capabilities, facilitating future updates to the knowledge graph.

[0053] The fault diagnosis agent is used to perform diagnostic reasoning on the event package based on the reference knowledge graph to obtain the target diagnosis result.

[0054] Optionally, in the step of performing diagnostic reasoning on the event package based on the reference knowledge graph to obtain the target diagnostic result, the fault diagnosis agent is specifically used for: E1. Determine a abnormal events corresponding to the event package; the a abnormal events include at least one of the following: threshold events and pattern events; a is a positive integer; E2. Determine the a signal nodes corresponding to the a abnormal events; E3. Traverse each of the a signal nodes according to the reference knowledge graph to obtain a set of fault-related nodes; E4. Determine a reference diagnostic results corresponding to the a fault-related node set according to the preset fault diagnosis model; E5. Calculate the confidence level of each of the a reference diagnostic results based on the fault case information to obtain a confidence levels; E6. Determine the confidence levels of the a confidence levels that are greater than the preset confidence threshold, and obtain at least one confidence level; E7. Based on the at least one confidence level, obtain at least one corresponding reference diagnostic result from the a reference diagnostic results to obtain the target diagnostic result.

[0055] In this embodiment of the application, the core information of the event package includes, but is not limited to: the total number of events, the event type, the associated signal ID, the event occurrence timestamp, and the signal context data, which are not specifically limited here.

[0056] In a specific embodiment, firstly, the event package is parsed to filter out all abnormal events contained within it, denoted as *a* abnormal events. The types of these abnormal events include at least one of threshold events and pattern events. A unique identifier can be established for each abnormal event, and it can be associated with its corresponding signal context data (such as signal timing segments 10 seconds before and after the event). Based on the associated signal ID corresponding to each abnormal event, node matching is performed in a reference knowledge graph to determine the signal node corresponding to each abnormal event, ultimately obtaining *a* signal nodes that correspond one-to-one with the *a* abnormal events. Specifically, using the associated signal ID of the abnormal event as the search keyword, a preset graph database query interface is called to search for signal nodes with the same signal ID attribute in the reference knowledge graph. If a matching node is found, that node is identified as the signal node corresponding to the abnormal event; if no matching node is found, the abnormal event is marked as "knowledge to be supplemented," and subsequent traversal of that abnormal event is skipped. Then, a bidirectional traversal operation is performed on each of the *a* signal nodes according to the reference knowledge graph to mine fault-related nodes that have direct or indirect associations with them, thereby obtaining a set of *a* fault-related nodes.

[0057] Then, the fault diagnosis agent can retrieve the preset fault diagnosis model, perform matching analysis on each fault-related node set, determine the reference diagnosis result corresponding to each fault-related node set, and finally obtain a reference diagnosis result that corresponds one-to-one with a fault-related node sets.

[0058] It should be noted that the fault diagnosis model is a combination of a knowledge graph-based rule-based reasoning model and a probabilistic causal model. Specifically, it matches node combinations from the fault-related node set with a fault diagnosis rule base developed by experts based on fault case information. If no clear rule is found for a node combination, the probabilistic causal model is invoked to calculate the probability of occurrence of each fault mode node in the fault-related node set. The probability calculation is based on statistical data from fault case information (e.g., the proportion of cases where "abnormal pump speed signal" corresponds to "pump jamming" is 85%).

[0059] Then, based on the fault case information, the confidence level of each of the *a* reference diagnostic results is calculated to obtain *a* confidence levels. This involves obtaining the rule matching degree, probabilistic causality value, and case matching degree for each reference diagnostic result. Then, the preset first weight, second weight, and third weight corresponding to the rule matching degree, probabilistic causality value, and case matching degree are obtained. A weighted calculation is performed based on the first weight, second weight, third weight, rule matching degree, probabilistic causality value, and case matching degree to obtain the confidence level for each reference diagnostic result. Specifically, the rule matching degree is 100% if the reference diagnostic result matches an explicit rule in the fault diagnosis rule base; otherwise, it is 0. The probabilistic causality value is the probability of fault occurrence calculated by the probabilistic causal model. The case matching degree is calculated by comparing the reasoning path of the current reference diagnostic result with the corresponding fault link in the fault case information. The confidence level ranges from 0-100%, with higher values ​​indicating stronger reliability of the reference diagnostic result.

[0060] Next, the 'a' confidence levels are filtered, retaining those with values ​​greater than or equal to a preset confidence threshold (the confidence threshold is determined by experts, with a default value of 80%, which can be adjusted according to diagnostic accuracy requirements), thus obtaining at least one satisfactory confidence level. Then, based on at least one confidence level, at least one corresponding reference diagnostic result from the 'a' reference diagnostic results is obtained to obtain the target diagnostic result. Specifically, if only one confidence level is obtained, its corresponding reference diagnostic result is directly used as the target diagnostic result; if multiple confidence levels are obtained, the corresponding reference diagnostic results are sorted from highest to lowest confidence level to generate a candidate fault list, and this candidate fault list is used as the target diagnostic result.

[0061] It is evident that combining rule-based reasoning with probabilistic causal models for dual evaluation improves the reliability of diagnostic results. Meanwhile, the screening mechanism based on confidence thresholds effectively reduces the risk of misdiagnosis.

[0062] Please see Figure 5 , Figure 5 This is a schematic diagram illustrating the composition of another vehicle fault diagnosis system based on a multi-agent system provided in this application embodiment. The system includes: a data acquisition agent, an event detection agent, a knowledge graph agent, a fault diagnosis agent, and an expert interaction agent. If the highest confidence level among the *a* confidence levels is lower than a preset confidence threshold, the expert interaction process of the expert interaction agent will be automatically triggered. The expert interaction agent can receive reasoning process data transmitted by the fault diagnosis agent and automatically integrate it to generate a structured, interpretable report. The expert interaction agent can receive feedback from experts regarding the interpretable report and then perform interactive operations based on this feedback. The interactive operations are as follows: Labeling new causal relationships: If key associations between missing nodes are found in the current reference knowledge graph, new causal relationship edges can be manually labeled. Add a new fault mode node: If it is determined that the current candidate fault list is missing a real fault, a new fault mode node can be added manually. Correcting existing relationship / node information: If incorrect node relationships (such as reversed causal relationships) or incorrect node attributes (such as incorrect fault level labeling) are found in the reference knowledge graph, you can directly modify them and submit them for review.

[0063] It is evident that the design of the expert interactive intelligent agent addresses the limitations of machine reasoning in complex fault scenarios and enables the dynamic evolution of the reference knowledge graph. On the one hand, the expert review mechanism for low-confidence reasoning results ensures the reliability of diagnostic results and avoids maintenance errors caused by misjudgments or omissions. On the other hand, the knowledge supplemented by experts is written back to the graph in real time, continuously enriching the system's diagnostic knowledge and improving the accuracy and efficiency of subsequent diagnoses. This eliminates the need for manual batch maintenance of diagnostic rules, significantly reducing system operation and maintenance costs, while also enhancing the system's adaptability to new fault modes and vehicle configuration iterations.

[0064] Optionally, in the step of traversing each of the a signal nodes according to the reference knowledge graph to obtain a set of fault-related nodes, the fault diagnosis agent is specifically used for: F1. Obtain the first node identifier of the first signal node; the first signal node is any one of the a signal nodes; F2. Determine the position of the first signal node in the reference knowledge graph based on the first node identifier to obtain the first position; F3. According to the preset traversal depth, taking the first position as the traversal starting point, traverse along the associated edges of the first signal node in the reference knowledge graph to obtain multiple first associated nodes; the node types of the multiple first associated nodes include: the component node and the fault mode node. F4. Based on the first signal node and the plurality of first associated nodes, determine the set of fault associated nodes corresponding to the first signal node in the set of a fault associated nodes.

[0065] In the embodiments of this application, In a specific embodiment, the fault diagnosis agent first extracts the first node identifier of the first signal node, which is a globally unique code for the signal node in the reference knowledge graph. The first signal node is any one of *a* signal nodes. By calling the query interface of the graph database to which the reference knowledge graph belongs, using the first node identifier as the search keyword, a node location operation is performed to obtain the first position. A precise first query instruction can be constructed based on the first node identifier. After receiving the first query instruction, the graph database quickly matches the target node using a preset node index and returns the unique position information of the target node in the graph (i.e., the first position).

[0066] Then, starting from the first position, a traversal operation is performed according to preset traversal parameters to uncover fault-related nodes that are directly or indirectly associated with the first signal node, resulting in multiple first associated nodes. The traversal parameters, including traversal depth and direction, can be retrieved first. The traversal depth is the preset maximum number of traversal layers (to avoid inefficiency due to an excessively large traversal range; the default is 3 layers, which can be dynamically adjusted according to the complexity of the vehicle model). The traversal direction is bidirectional (traversing upstream inducing nodes and downstream influencing nodes simultaneously along the associated edges to ensure no associations are missed). During the traversal along the associated edges, causal relationship edges (core associated edges, directly reflecting the fault propagation logic) in the reference knowledge graph can be traversed first, followed by dependency relationship edges (auxiliary associated edges, indirectly reflecting the supporting relationships of components / signals). Upstream traversal can trace potential inducing nodes of the abnormal signal, while downstream traversal can trace chain-effect nodes that the abnormal signal may trigger. Then, nodes of type component node and fault mode node are selected (excluding non-fault core nodes such as control logic nodes and irrelevant signal nodes), and finally the multiple first associated nodes are obtained.

[0067] Finally, the first signal node is structurally integrated with multiple first associated nodes to form a fault associated node set corresponding to the first signal node. It should be noted that the fault associated node set also includes the type of associated edges (causal relationship edges / dependency relationship edges) and the associated paths between each node, which are not specifically limited here.

[0068] It is evident that the positioning method based on unified signal identification ensures the accuracy of node matching; the combination of bidirectional traversal and depth control balances traversal efficiency and correlation coverage; the filtering rules for component nodes and fault mode nodes eliminate redundant information and improve the efficiency of subsequent diagnostic reasoning; and the structured node set encapsulation provides a standardized data format for batch matching of fault diagnosis models.

[0069] For easier understanding, please refer to Figure 6 , Figure 6This diagram illustrates an application scenario of a multi-agent-based vehicle fault diagnosis system provided in this application. The target vehicle generates operational data (such as sensor signals, control commands, and fault codes) and sends this data to the multi-agent-based vehicle fault diagnosis system. Upon receiving the operational data, the system, through the collaboration of multiple internal agents (such as data acquisition, event detection, knowledge graph, and fault diagnosis agents), completes data processing, anomaly detection, and reasoning analysis, ultimately generating a diagnostic result. Finally, the diagnostic result is fed back to the target user, who can be a repair technician, expert, or vehicle owner, allowing them to understand the vehicle's fault condition.

[0070] The following is combined Figure 7 The electronic devices in the embodiments of this application will be described. Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application, such as... Figure 7 As shown, the electronic device includes one or more processors, a memory, a communication interface, and one or more programs. The processor is connected to the memory and the communication interface via an internal communication bus.

[0071] The processor can be used for: Collect target vehicle operation data; Event detection is performed on the target vehicle's operating data to obtain event packets; Obtain the vehicle model information corresponding to the target vehicle; construct a reference knowledge graph corresponding to the target vehicle based on the vehicle model information; The event package is diagnosed and reasoned based on the reference knowledge graph to obtain the target diagnosis result.

[0072] The one or more programs are stored in the aforementioned memory and configured to be executed by the aforementioned processor, and the one or more programs include instructions for performing any of the steps in the above embodiments.

[0073] The processor can be a central processing unit (CPU), a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, cells, and circuits described in conjunction with the disclosure of this application. The processor can also be a combination that implements computational functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, etc. The communication unit can be a communication interface, transceiver, transceiver circuit, etc., and the storage unit can be a memory.

[0074] The memory can be volatile or non-volatile, or a combination of both. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory can be random access memory (RAM), used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0075] It is understood that the electronic device may include more or fewer structural elements than those shown in the block diagram above, such as a power module, physical buttons, a Wi-Fi module, a speaker, a Bluetooth module, sensors, a display module, etc., without limitation. It is understood that the electronic device may incorporate elements such as... Figure 1 The system architecture described above.

[0076] The following is combined Figure 8 This application describes a vehicle fault diagnosis method based on multi-agent technology. Figure 8 This is a flowchart illustrating a vehicle fault diagnosis method based on multi-agent technology provided in an embodiment of this application, specifically including the following steps: S1. Collect target vehicle operation data; S2. Perform event detection on the target vehicle's operating data to obtain an event packet; S3. Obtain the vehicle model information corresponding to the target vehicle; construct a reference knowledge graph corresponding to the target vehicle based on the vehicle model information; S4. Perform diagnostic reasoning on the event package based on the reference knowledge graph to obtain the target diagnostic result.

[0077] In one possible embodiment, the collection of target vehicle operation data includes: Obtain multiple vehicle communication buses corresponding to the target vehicle; Data is collected from the target vehicle through the multiple vehicle communication buses to obtain multiple initial vehicle operation data; each vehicle communication bus corresponds to one initial vehicle operation data. Each of the multiple initial vehicle operation data is preprocessed to obtain multiple reference vehicle operation data. The multiple reference vehicle operation data are encapsulated according to a preset signal flow structure to obtain the target vehicle operation data; the target vehicle operation data includes multiple signal flows, each signal flow corresponding to one reference vehicle operation data.

[0078] In one possible embodiment, the step of performing event detection on the target vehicle's operating data to obtain an event packet includes: Determine the abnormal pattern feature library corresponding to the target vehicle's operating data; Determine multiple signal threshold ranges corresponding to the multiple signal streams in the target vehicle's operating data; each signal stream corresponds to one signal threshold range. If the signal value of at least one first signal stream in the plurality of signal streams exceeds its corresponding signal threshold range, then at least one threshold event is generated based on the at least one first signal stream. If at least one of the multiple signal streams has a signal feature that matches the abnormal pattern feature library, then at least one pattern event is generated based on the at least one second signal stream. The event package is determined based on the at least one threshold event and / or the at least one pattern event.

[0079] In one possible embodiment, the step of determining if the signal characteristics of at least one second signal stream among the plurality of signal streams match the abnormal pattern feature library includes: According to a preset sliding window mechanism, continuous time segments of each of the multiple signal streams are extracted to obtain multiple continuous time segments; Extract features of multiple reference signals corresponding to the multiple time segments; Obtain the multiple abnormal signal features corresponding to the multiple reference signal features in the abnormal pattern feature library; Based on a preset dynamic time warping algorithm, the matching degree between the multiple reference signal features and the multiple abnormal signal features is calculated to obtain multiple matching degrees; Obtain at least one matching degree from among the plurality of matching degrees that are greater than a preset matching degree threshold; The at least one signal stream corresponding to the at least one matching degree among the plurality of signal streams is determined to be the at least one second signal stream.

[0080] In one possible embodiment, constructing the reference knowledge graph corresponding to the target vehicle based on the vehicle model information includes: Based on the vehicle model information, multiple nodes corresponding to the target vehicle are determined; the node types of the multiple nodes include: component nodes, signal nodes, and fault mode nodes; Obtain the fault case information corresponding to the vehicle model information; Based on the fault case information, determine the edge relationship between any two nodes among the plurality of nodes to obtain an edge set; The reference knowledge graph is constructed based on the plurality of nodes and the set of edges.

[0081] In one possible embodiment, the step of performing diagnostic reasoning on the event package based on the reference knowledge graph to obtain a target diagnostic result includes: Identify a abnormal events corresponding to the event package; the a abnormal events include at least one of the following: threshold events and pattern events; a is a positive integer; Identify the a signal nodes corresponding to the a abnormal events; Based on the reference knowledge graph, each of the a signal nodes is traversed to obtain a set of a fault-related nodes. Based on the preset fault diagnosis model, determine a reference diagnosis results corresponding to the a fault-related node set; Based on the fault case information, the confidence level of each of the a reference diagnostic results is calculated to obtain a confidence levels; Determine the confidence levels of the a confidence levels that are greater than a preset confidence threshold, and obtain at least one confidence level; Based on the at least one confidence level, at least one corresponding reference diagnostic result is obtained from the a reference diagnostic results to obtain the target diagnostic result.

[0082] In one possible embodiment, the step of traversing each of the a signal nodes according to the reference knowledge graph to obtain a set of fault-related nodes includes: Obtain the first node identifier of the first signal node; the first signal node is any one of the a signal nodes; The position of the first signal node in the reference knowledge graph is determined based on the first node identifier, thus obtaining the first position; According to the preset traversal depth, starting from the first position, the traversal is performed along the associated edges of the first signal node in the reference knowledge graph to obtain multiple first associated nodes; the node types of the multiple first associated nodes include: the component node and the fault mode node. Based on the first signal node and the plurality of first associated nodes, determine the set of fault associated nodes corresponding to the first signal node in the set of a fault associated nodes.

[0083] As can be seen, by collecting multi-dimensional data covering fault codes, freeze frames, real-time sensor streams, control commands, and actuator feedback, the system covers the entire vehicle operation scenario. Simultaneously, through preprocessing and encapsulation operations, it achieves time synchronization, anomaly removal, and format unification of multi-source heterogeneous data, avoiding diagnostic biases caused by data clutter and improving the efficiency and accuracy of subsequent event detection and inference. Based on threshold and pattern dual detection logic, it identifies abnormal events, capturing both explicit faults with excessive signal values ​​and implicit fault precursors with abnormal signal timing patterns. Compared to traditional single threshold detection, the false negative rate is significantly reduced. Furthermore, the event package integrates multiple types of anomaly information, providing complete anomaly input for diagnostic inference. Based on vehicle model information, it constructs a target vehicle-specific reference knowledge graph, overcoming the universality limitations of traditional fixed rule bases and accurately adapting to the hardware configuration, electronic and electrical architecture, and fault mode characteristics of different vehicle models. Based on a dedicated knowledge graph, targeted reasoning is performed on event packages, and confidence is quantitatively assessed by combining fault case information to screen highly reliable diagnostic results, effectively reducing the risk of misjudgment. At the same time, the knowledge graph can be dynamically updated with new fault cases, eliminating the need for manual maintenance of rules one by one. It adapts to vehicle configuration iterations and the emergence of new fault modes, significantly reducing system maintenance costs and improving the response speed of technology iteration.

[0084] This application also provides a computer-readable storage medium storing a computer program for electronic data interchange that causes a computer to perform some or all of the steps of the multi-agent-based vehicle fault diagnosis method described in the above embodiments, wherein the computer includes an electronic device.

[0085] This application also provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program operable to cause a computer to perform some or all of the steps of the multi-agent-based vehicle fault diagnosis method described in the above embodiments. The computer program product can be a software installation package, and the computer includes an electronic device.

[0086] It should be noted that, for the sake of simplicity, the above embodiments are all described as a series of actions. Those skilled in the art should understand that this application is not limited to the described order of actions, as some steps in the embodiments of this application can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions, steps, modules, or units involved are not necessarily essential to the embodiments of this application.

[0087] In the above embodiments, the descriptions of each embodiment in this application have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0088] Those skilled in the art will understand that implementing all or part of the processes in the above embodiments can be accomplished by a computer program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

[0089] The steps of the methods or algorithms described in the embodiments of this application can be implemented in hardware or by a processor executing software instructions. The software instructions can consist of corresponding software modules, which can be stored in RAM, flash memory, ROM, EPROM, electrically erasable programmable read-only memory (EEPROM), registers, hard disk, portable hard disk, read-only optical disk (CD-ROM), or any other form of storage medium well known in the art. An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and storage medium can reside in an ASIC. Furthermore, the ASIC can reside in a terminal device or management device. Alternatively, the processor and storage medium can exist as discrete components in the terminal device or management device.

[0090] Those skilled in the art will recognize that, in one or more of the examples above, the functions described in the embodiments of this application can be implemented, in whole or in part, by software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. This computer program product includes one or more computer instructions. When these computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., digital video discs (DVDs)), or semiconductor media (e.g., solid-state disks (SSDs)).

[0091] The modules / units included in the various devices and products described in the above embodiments can be software modules / units, hardware modules / units, or a combination of both. For example, for devices and products applied to or integrated into a chip, all modules / units can be implemented using hardware methods such as circuits, or at least some modules / units can be implemented using software programs that run on a processor integrated within the chip, while the remaining (if any) modules / units can be implemented using hardware methods such as circuits. For devices and products applied to or integrated into a chip module, all modules / units can be implemented using hardware methods such as circuits. Different modules / units can be located in the same component (e.g., chip, circuit module, etc.) or different components of the chip module, or at least some modules / units can be implemented using hardware methods such as circuits. The implementation is achieved through a software program that runs on the processor integrated within the chip module. The remaining modules / units (if any) can be implemented using hardware methods such as circuits. For various devices and products applied to or integrated into terminal equipment, each of their modules / units can be implemented using hardware methods such as circuits. Different modules / units can be located in the same component (e.g., chip, circuit module, etc.) or different components within the terminal equipment. Alternatively, at least some modules / units can be implemented through a software program that runs on the processor integrated within the terminal equipment, while the remaining modules / units (if any) can be implemented using hardware methods such as circuits.

[0092] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the embodiments of this application. It should be understood that the above descriptions are merely specific embodiments of the embodiments of this application and are not intended to limit the protection scope of the embodiments of this application. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solutions of the embodiments of this application should be included within the protection scope of the embodiments of this application.

Claims

1. A vehicle fault diagnosis system based on multi-agent technology, characterized in that, The system includes: a data acquisition agent, an event detection agent, a knowledge graph agent, and a fault diagnosis agent, wherein: The data acquisition intelligent agent is used to collect target vehicle operation data of the target vehicle. The event detection agent is used to perform event detection on the target vehicle's operating data to obtain an event packet; The knowledge graph agent is used to obtain vehicle model information corresponding to the target vehicle; and to construct a reference knowledge graph corresponding to the target vehicle based on the vehicle model information. The fault diagnosis agent is used to perform diagnostic reasoning on the event package based on the reference knowledge graph to obtain the target diagnosis result.

2. The system as described in claim 1, characterized in that, Regarding the collection of target vehicle operation data, the data collection agent is specifically used for: Obtain multiple vehicle communication buses corresponding to the target vehicle; Data is collected from the target vehicle through the multiple vehicle communication buses to obtain multiple initial vehicle operation data; each vehicle communication bus corresponds to one initial vehicle operation data. Each of the multiple initial vehicle operation data is preprocessed to obtain multiple reference vehicle operation data. The multiple reference vehicle operation data are encapsulated according to a preset signal flow structure to obtain the target vehicle operation data; the target vehicle operation data includes multiple signal flows, each signal flow corresponding to one reference vehicle operation data.

3. The system as described in claim 2, characterized in that, In the process of detecting events in the target vehicle's operating data to obtain event packets, the event detection agent is specifically used for: Determine the abnormal pattern feature library corresponding to the target vehicle's operating data; Determine multiple signal threshold ranges corresponding to the multiple signal streams in the target vehicle's operating data; each signal stream corresponds to one signal threshold range. If the signal value of at least one first signal stream in the plurality of signal streams exceeds its corresponding signal threshold range, then at least one threshold event is generated based on the at least one first signal stream. If at least one of the multiple signal streams has a signal feature that matches the abnormal pattern feature library, then at least one pattern event is generated based on the at least one second signal stream. The event package is determined based on the at least one threshold event and / or the at least one pattern event.

4. The system as described in claim 3, characterized in that, In terms of the condition that at least one second signal stream among the plurality of signal streams matches the abnormal pattern feature library, the event detection agent is specifically used for: According to a preset sliding window mechanism, continuous time segments of each of the multiple signal streams are extracted to obtain multiple continuous time segments; Extract features of multiple reference signals corresponding to the multiple time segments; Obtain the multiple abnormal signal features corresponding to the multiple reference signal features in the abnormal pattern feature library; Based on a preset dynamic time warping algorithm, the matching degree between the multiple reference signal features and the multiple abnormal signal features is calculated to obtain multiple matching degrees; Obtain at least one matching degree from among the plurality of matching degrees that are greater than a preset matching degree threshold; The at least one signal stream corresponding to the at least one matching degree among the plurality of signal streams is determined to be the at least one second signal stream.

5. The system according to any one of claims 1-4, characterized in that, In constructing the reference knowledge graph corresponding to the target vehicle based on the vehicle model information, the knowledge graph agent is specifically used for: Based on the vehicle model information, determine multiple nodes corresponding to the target vehicle; The node types of the multiple nodes include: component nodes, signal nodes, and fault mode nodes; Obtain the fault case information corresponding to the vehicle model information; Based on the fault case information, determine the edge relationship between any two nodes among the plurality of nodes to obtain an edge set; The reference knowledge graph is constructed based on the plurality of nodes and the set of edges.

6. The system as described in claim 5, characterized in that, In terms of performing diagnostic reasoning on the event package based on the reference knowledge graph to obtain the target diagnostic result, the fault diagnosis agent is specifically used for: Identify a abnormal events corresponding to the event package; the a abnormal events include at least one of the following: threshold events and pattern events; a is a positive integer; Identify the a signal nodes corresponding to the a abnormal events; Based on the reference knowledge graph, each of the a signal nodes is traversed to obtain a set of a fault-related nodes. Based on the preset fault diagnosis model, determine a reference diagnosis results corresponding to the a fault-related node set; Based on the fault case information, the confidence level of each of the a reference diagnostic results is calculated to obtain a confidence levels; Determine the confidence levels of the a confidence levels that are greater than a preset confidence threshold, and obtain at least one confidence level; Based on the at least one confidence level, at least one corresponding reference diagnostic result is obtained from the a reference diagnostic results to obtain the target diagnostic result.

7. The system as described in claim 6, characterized in that, In the process of traversing each of the a signal nodes according to the reference knowledge graph to obtain a set of a fault-related nodes, the fault diagnosis agent is specifically used for: Obtain the first node identifier of the first signal node; the first signal node is any one of the a signal nodes; The position of the first signal node in the reference knowledge graph is determined based on the first node identifier, thus obtaining the first position; According to the preset traversal depth, taking the first position as the traversal starting point, traversing along the associated edges of the first signal node in the reference knowledge graph, multiple first associated nodes are obtained. The node types of the plurality of first associated nodes include: the component nodes and the fault mode nodes; Based on the first signal node and the plurality of first associated nodes, determine the set of fault associated nodes corresponding to the first signal node in the set of a fault associated nodes.

8. A vehicle fault diagnosis method based on multi-agent systems, characterized in that, The method includes: Collect target vehicle operation data; Event detection is performed on the target vehicle's operating data to obtain event packets; Obtain the vehicle model information corresponding to the target vehicle; construct a reference knowledge graph corresponding to the target vehicle based on the vehicle model information; The event package is diagnosed and reasoned based on the reference knowledge graph to obtain the target diagnosis result.

9. An electronic device, characterized in that, include: Processor, memory, communication interface, and one or more programs; The one or more programs are stored in the memory and configured to be executed by the processor, the programs including instructions for performing the steps in the method of claim 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, the computer program including program instructions that, when executed by a processor, cause the processor to perform the method as described in claim 8.

Images