Vehicle health management system based on knowledge graph technology

Abstract

This invention relates to a vehicle health management system based on knowledge graph technology, belonging to the field of intelligent vehicle management. It includes a knowledge graph model, a data processing module, an intelligent search module, a health assessment module, a fault strategy module, and a maintenance strategy module. The data processing module monitors and processes the functions of various vehicle devices. The intelligent search module matches the device's attribute information and functional data with the knowledge graph model, aligning data with physical objects and accurately locating the functional position of devices with abnormal data. The health assessment module determines whether a device can continue to operate normally and assesses its health status. The fault strategy and maintenance strategy modules execute corresponding actions based on the functional assessment values. This invention enables comprehensive, multi-element, and full-process dynamic perception, real-time monitoring, data integration, intelligent control, and self-decision-making for all devices during vehicle use.

Description

Technical Field

[0001] This invention relates to the fields of knowledge graph technology and intelligent vehicle management, and in particular to a vehicle health management system based on knowledge graph technology. Background Technology

[0002] Artificial intelligence (AI) technology is a core technology and development direction for the future of vehicle engineering. Knowledge graph technology, as an important component of AI, is a key technology for building an intelligent vehicle health management system capable of self-perception, self-learning, self-decision-making, and self-execution. Vehicle health management is a comprehensive concept and technical system designed to ensure vehicles are always in good working condition and fully utilize their performance, which is especially important for armored equipment.

[0003] Currently, vehicle health management in the field of automotive engineering is relatively simple. For example, vehicle maintenance intervals are calculated based on mileage or the time since the last maintenance. This method only allows for static analysis of the operational status of various vehicle components, potentially leading to over-maintenance, increased costs, and wasted resources. Delayed maintenance can also severely impact vehicle lifespan or increase failure rates. Furthermore, while current vehicle designs incorporate sensors in critical components, these sensors may provide warnings if a malfunction occurs during operation, but they cannot pinpoint the fault location or provide solutions. Repairs must rely on traditional manual methods and experience, which is inefficient and makes accurate fault identification and location difficult, increasing repair time and costs. Additionally, a review of recent traffic accidents reveals a significant proportion caused by vehicle malfunctions during operation. This is primarily because current vehicle systems lack comprehensive monitoring and evaluation of the operational status of various components, making it impossible to predict their condition and intervene proactively. If a vehicle malfunctions during operation, it can range from affecting traffic safety to causing fatal accidents. Improvements are urgently needed. Summary of the Invention

[0004] The purpose of this invention is to provide a vehicle health management system based on knowledge graph technology, which solves the problems of existing vehicle health management methods, such as overall disorganization, incomplete driving data monitoring, inaccurate fault type localization, inefficient maintenance and repair, and lack of early warning for system component health. This invention can achieve comprehensive, all-element, and all-stage dynamic perception, real-time monitoring, data integration, intelligent control, and self-decision-making for all devices during vehicle use.

[0005] The above-mentioned objective of the present invention is achieved through the following technical solution:

[0006] The vehicle health management system based on knowledge graph technology includes a knowledge graph model, a data processing module, an intelligent search module, a health assessment module, a fault strategy module, and a maintenance strategy module. The knowledge graph model is the knowledge foundation and underlying design of this invention, comprising three sub-models: knowledge input, knowledge fusion, and knowledge output.

[0007] In the knowledge input sub-model, two data model databases are established: one for the original attributes of each device and the other for functional information. First, based on the original attribute parameters of each device, an original attribute database for each device in the vehicle is established. Then, the working status data of each device in the vehicle under different usage stages and operating conditions are monitored to establish a functional information database for each device under all usage stages and operating conditions. The data in this database is divided into three data types: structured data, semi-structured data, and unstructured data.

[0008] In the knowledge fusion sub-model, data is aligned with entities through logical computation, and an entity data model is constructed. The device's health status is inferred from the data information and updated in real time. Structured data refers to data with a clear, predefined data model and following a consistent order. Data from the device's original attribute database and functional information database can be directly integrated, allowing direct location of the specific device and its operating status. Unstructured data refers to data without a predefined data model, with irregular or incomplete data structures. Semi-structured data falls between structured and unstructured data, possessing certain structured characteristics but not fully conforming to them. Since these two types of data cannot directly locate specific devices and their operating status, knowledge extraction is required. This is mainly achieved through entity recognition, relationship understanding, attribute filtering, and data formatting to extract knowledge points from the data and store them in a knowledge base in a certain form. This enhances the usability and importance of the data information, thereby enabling entity annotation of semi-structured and unstructured data.

[0009] Furthermore, by integrating the original attribute database and functional information database of the devices with the semi-structured and unstructured entity-annotated data, logical calculations are performed to ensure that each data knowledge point is aligned one-to-one with the relevant features of the device entity. This means that different data combinations can accurately identify each device in the vehicle and its operational status. Next, functional data information is labeled, and different functional information data are assigned functional values ​​to achieve quality assessment of the relevant devices. Additionally, real-time updates based on the health status data of relevant devices during vehicle operation are necessary to better reflect the quality and consumption of these devices.

[0010] In the knowledge output sub-model, the data network from the knowledge fusion stage is interwoven into a knowledge graph with high relevance, strong correspondence, wide coverage, and accurate positioning. Knowledge is then output through graph retrieval.

[0011] The data processing module mainly comprises three sub-modules: system monitoring, data acquisition, and data comparison. It is primarily used for functional monitoring and data processing of various vehicle devices, providing instructions for executing the next action. Specifically, the system monitoring sub-module acquires attribute information of each vehicle device and monitors its functional status. Its key features are full vehicle coverage, full information coverage, and full functional coverage, providing the logical foundation for subsequent command actions. The data acquisition sub-module collects raw attribute information and functional status data of each device using high-precision, highly durable data sensors deployed throughout the vehicle. Its key features are the speed, accuracy, and comprehensiveness of raw attribute information and functional status data acquisition, providing the data foundation for subsequent model calculations. The data comparison sub-module compares and extracts data from the original attribute information and functional data of each device through data visualization processing or simulation calculations to obtain information models of each device. Based on this, it calculates the corresponding functional anomaly curve for each device, determines whether each device is functionally abnormal based on these curves, and further executes subsequent logical actions.

[0012] To determine the functional status of each device, a functional anomaly identification model needs to be established based on the information model of each device. Its expression is as follows:

[0013] ;

[0014] In the formula: n is the input data, which is the functional monitoring data of the device; FB(n) is the abnormal value of the function;

[0015] When the output value FB(n) is 1, it is determined that the detection device is functioning normally and no corresponding operation is performed.

[0016] When the output value FB(n) is 0, it is determined that the detected device is malfunctioning, and the system will continue to execute logical actions in the intelligent search module.

[0017] The intelligent search module mainly comprises three sub-modules: data extraction, data classification, and graph search. Its primary function is to match the device's attribute information and functional data with a knowledge graph model, aligning data with the physical object and accurately locating the functional location of devices with abnormal data. Specifically, the data extraction sub-module extracts abnormal functional data through logical calculations and data comparisons, and locates the abnormal data to the corresponding device using attribute information, achieving precise data-physical alignment. The data classification sub-module categorizes abnormal device data into direct abnormal data and verification abnormal data. Direct abnormal data can be directly aligned with device malfunctions, indicating a functional abnormality. Verification abnormal data requires verification of the device's function, with verification results including both equipment-related and non-equipment-related causes, further determining whether the device is functionally abnormal. The graph search sub-module primarily searches, matches, and retrieves relevant device information data within the established vehicle knowledge graph model, accurately locating the functional abnormality of the corresponding device, providing precise information support for the next step of device health status assessment.

[0018] The health assessment module mainly consists of two sub-modules: correlation calculation and anomaly evaluation. The correlation calculation sub-module performs correlation calculations between the abnormal data collected from the abnormal device and the corresponding data in the knowledge graph to further determine whether the device is abnormal and whether it can continue to operate normally. The anomaly evaluation sub-module further evaluates the abnormal data identified in the correlation calculation sub-module to assess the device's health status in order to proceed with the next step.

[0019] In the entire health assessment module, to evaluate the function of the corresponding device, the device data is first fitted to obtain a function curve function, denoted as GY(t), where t is time. Based on this function, a function assessment model is established, the expression of which is:

[0020]

[0021] In the formula: FA is the functional evaluation value; δt is the coefficient function, and the coefficient function... ; t0 is the starting time, which is the time corresponding to the closest t0 to the current time; dt represents the integration over time.

[0022] When the functional evaluation value is 0.8≤FA≤1, the evaluation function is normal, indicating that the corresponding device has low wear and tear and the device function is normal.

[0023] When the functional evaluation value is 0.2 < FA < 0.8, an abnormal evaluation function is indicated, which means that the corresponding device is worn out, but does not affect the function. Maintenance can be performed as needed.

[0024] When the functional evaluation value is 0 < FA ≤ 0.2, a functional failure warning is issued, indicating that the wear and tear of the corresponding device is increasing and has gradually affected the normal operation and function of the device. The device should be inspected and maintained immediately. If the conditions for inspection or maintenance are not available, the vehicle may be allowed to travel to the vehicle inspection and maintenance location for a short period of time to avoid failure during the vehicle's journey.

[0025] When the functional evaluation value FA=0, the evaluation function is faulty, indicating that the device is completely unable to work and has lost its corresponding function. It must be repaired; otherwise, the vehicle cannot be started or driven.

[0026] The fault strategy module and maintenance strategy module mainly execute corresponding actions based on the functional evaluation value. If the functional evaluation value is 0.8 ≤ FA ≤ 1, the functional evaluation is normal, and no corresponding operation is performed. If the functional evaluation value is 0.2 < FA < 0.8, a functional evaluation abnormality is indicated, and the system will provide a maintenance strategy based on the abnormal location of the corresponding device. The user can then operate as needed. If the functional evaluation value is 0 < FA ≤ 0.2, a functional evaluation fault warning is issued, and the system will provide a maintenance strategy based on the abnormal location of the corresponding device. The user must take immediate action for maintenance. If inspection or maintenance conditions are not available, the vehicle can be driven to a vehicle inspection and maintenance location for a short period of time to prevent malfunctions during vehicle operation. If the functional evaluation value FA = 0, a functional fault is evaluated, and the system will provide a repair strategy based on the fault status of the corresponding device. The user must repair the corresponding device immediately; otherwise, the vehicle cannot be started or driven.

[0027] After the corresponding device is repaired or maintained, the system will re-monitor its working status and enter a new cycle.

[0028] The beneficial effects of this invention are as follows: Compared with the prior art, this invention has the following advantages:

[0029] I. Innovative application of knowledge graph technology in the field of vehicle engineering. Knowledge graph technology is an important component of artificial intelligence. Embedding knowledge graphs into vehicle management systems can greatly improve the vehicle's self-perception, self-learning, self-decision-making, and self-execution management capabilities, making vehicle management more intelligent, faster, more accurate, and more effective.

[0030] Second, it innovatively explores a new concept of vehicle health management in the field of vehicle engineering. Vehicle health management is an innovative concept and comprehensive technical system based on big data analysis. It can monitor the "health" status of various devices in a vehicle in real time, allowing users to understand the vehicle's working status in real time, which can improve users' sense of security when driving. If a device experiences abnormal fluctuations in operation, the system can also provide users with effective solutions, thereby realizing more effective human-computer interaction.

[0031] Third, an innovative device function evaluation model was established. This model is the core basis for the vehicle health management system to execute operation commands. Through this model, the working status of vehicle devices is innovatively divided into four status zones: normal function zone, abnormal indication zone, fault warning zone, and functional failure zone, which effectively solves the problem of unclear wear and tear throughout the device's entire life cycle.

[0032] Fourth, an innovative vehicle fault prevention mechanism has been established. By integrating knowledge graphs with the vehicle management system, a vehicle health management system has been built. This system can monitor the working status and functional information of various vehicle devices in real time through mathematical models, logical calculations, and algorithmic judgments. After intelligent calculation and processing of the data, it can provide users with the usage status of the corresponding devices and offer effective handling strategies, thus better preventing vehicle faults and avoiding accidents. Attached Figure Description

[0033] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate the invention and are used to explain it, but do not constitute an undue limitation of the invention.

[0034] Figure 1 A principle block diagram for constructing the knowledge graph model of this invention;

[0035] Figure 2 This is a schematic diagram of the vehicle health management system of the present invention. Detailed Implementation

[0036] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0037] See Figure 1 and Figure 2 As shown, the vehicle health management system based on knowledge graph technology of the present invention includes "one graph and five modules", namely, knowledge graph model, data processing module, intelligent search module, health assessment module, fault strategy module, and maintenance strategy module.

[0038] See Figure 1 As shown, the knowledge graph model is the knowledge foundation and underlying design of this invention, including three sub-models: knowledge input, knowledge fusion, and knowledge output.

[0039] The knowledge input sub-model requires the establishment of two data model databases: one for the original attributes of each device and the other for its functional information. First, based on the original attribute parameters of each device, including name, type, parameters, function, and installation location, a database of the original attributes of each vehicle device is established. This database is then transformed into a digital, three-dimensional data model using 3D visualization technology. Next, functional monitoring is performed on the operating status data of each device in the vehicle under different usage stages and conditions. This includes data such as the air filter's filtration effect or drag coefficient, the oil filter's filtration effect or oil pressure, and the amount or temperature of coolant. This yields a data model database of functional information for each device under all usage stages and conditions, which is then displayed within the digital, three-dimensional data model.

[0040] The knowledge fusion sub-model aligns data with entities through logical calculations, constructs an entity data model, infers the device's health status based on data information, and updates it in real time. Based on the device attribute information model library and functional data model library, structured, semi-structured, and unstructured data are fused to establish a comprehensive digital three-dimensional data model. Alignment is performed one-to-one based on the relationship between information and data and entities. For example, by detecting the drag coefficient, it can be aligned with the working status of the air filter. If the drag coefficient is outside the normal operating range, it indicates that the air filter element is clogged or damaged. A one-to-one correspondence between the drag coefficient and the wear and tear of the air filter element is established; therefore, the health status of the air filter element can be clearly determined by the drag coefficient.

[0041] The knowledge output sub-model interweaves the data network from the knowledge fusion stage into a knowledge graph with high correlation, strong correspondence, wide coverage, and accurate positioning, so that the system can retrieve and search it.

[0042] The data processing module mainly comprises three sub-modules: system monitoring, data acquisition, and data comparison. It is primarily used for functional monitoring and data processing of various vehicle devices, providing instructions for executing the next action. The system monitoring sub-module acquires attribute information of each vehicle device, such as device name, type, parameters, function, and installation location, and monitors their functional status, including drag coefficient, oil pressure, oil temperature, coolant temperature, engine speed, and exhaust temperature. Its key features are comprehensive vehicle coverage, comprehensive information coverage, and comprehensive functional coverage, providing the logical foundation for subsequent instructions and actions. The data acquisition sub-module collects raw attribute information and functional status data of each device through high-precision, highly durable data sensors deployed throughout the vehicle. Its key features are the speed, accuracy, and comprehensiveness of the raw attribute information and functional status data acquisition. This provides a data foundation for subsequent model calculations. The data comparison submodule mainly compares and extracts the original attribute information and functional data of each device through data visualization processing or simulation calculation to obtain the information model of each device. Based on this, the corresponding functional abnormality curve of each device is calculated. Based on this functional abnormality curve, it is determined whether each device is functionally abnormal, and further executes subsequent logical actions. For example, the normal operating range of the oil pressure of the BF8L413FC diesel engine of a certain type of vehicle is 382-450KPa, while the oil pressure measured at a certain moment according to the functional abnormality curve is 400KPa. After data comparison, it can be found that the oil pressure is normal, indicating that the device is functioning normally.

[0043] To enable the system to determine the functional status of each device and facilitate logic execution, a functional anomaly identification model needs to be established based on the information model of each device. Its expression is as follows:

[0044] ;

[0045] In the formula: n is the input data, which is the functional monitoring data of the corresponding device; FB(n) is the abnormal value of the function;

[0046] When the output value FB(n) is 1, it is determined that the detection device is functioning normally and no corresponding operation is performed.

[0047] When the output value FB(n) is 0, it is determined that the detected device is malfunctioning, and the system will continue to execute logical actions in the intelligent search module.

[0048] The intelligent search module mainly includes three sub-modules: data extraction, data classification, and graph search. Its main function is to match the device's attribute information and functional data with a knowledge graph model, achieving data-to-physical alignment and accurately locating the functional position of devices with abnormal data. Specifically, the data extraction sub-module extracts abnormal functional data through logical calculations and data comparisons, and locates the abnormal data to the corresponding device using attribute information, achieving precise data-to-physical alignment. For example, after the above actions are completed, if a data point is extracted as the drag coefficient value, since the drag indicator is installed on the air filter, this data can be directly located on the air filter. The data classification sub-module divides abnormal device data into direct abnormal data and verification abnormal data. Direct abnormal data can be directly aligned with device malfunctions, indicating a functional abnormality. Verification abnormal data requires verification of the device's function, and the verification results include both equipment-related and non-equipment-related causes, further determining whether the device is functionally abnormal. For example, after the above actions are completed, if a data point is extracted as the oil pressure value, the oil... Fluctuations in oil pressure values ​​are related to many factors, such as whether the filter is clogged or the ambient temperature is low. If the filter is clogged, it's a device-related issue. However, if the ambient temperature is low and the vehicle has just started, the oil pressure may increase, which is not a device-related issue. The knowledge graph search submodule mainly searches, matches, and retrieves relevant device information data within the established vehicle knowledge graph model. It accurately matches and locates the functional abnormalities of corresponding devices, providing precise information support for the next step of device health status assessment. For example, if after the above actions are completed, a data point is extracted indicating high oil pressure, adequate oil quantity, normal engine temperature, and normal other parameters, a search in the knowledge graph can directly identify the cause as a clogged oil filter element and provide corresponding solutions.

[0049] The health assessment module mainly includes two sub-modules: correlation calculation and anomaly evaluation. The correlation calculation sub-module performs correlation calculations between the abnormal data collected from the abnormal device and the corresponding data in the knowledge graph to further determine whether the device is abnormal and whether it can continue to operate normally. The anomaly evaluation sub-module further evaluates the abnormal data identified in the correlation calculation sub-module to assess the device's health status in order to proceed with the next step.

[0050] In the entire health assessment module, to evaluate the function of the corresponding device, the device data is first fitted to obtain a function curve function, denoted as GY(t), where t is time. Based on this function, a function assessment model is established, the expression of which is:

[0051]

[0052] In the formula: FA is the functional evaluation value; δt is the coefficient function, and the coefficient function... ; t0 is the starting time, which is the time corresponding to the closest t0 to the current time; dt represents the integration over time.

[0053] When the functional evaluation value is 0.8≤FA≤1, the evaluation function is normal, indicating that the corresponding device has low wear and tear and the device function is normal.

[0054] When the functional evaluation value is 0.2 < FA < 0.8, an abnormal evaluation function is indicated, which means that the corresponding device is worn out, but does not affect the function. Maintenance can be performed as needed.

[0055] When the functional evaluation value is 0 < FA ≤ 0.2, a functional failure warning is issued, indicating that the wear and tear of the corresponding device is increasing and has gradually affected the normal operation and function of the device. The device should be inspected and maintained immediately. If the conditions for inspection or maintenance are not available, the vehicle may be allowed to travel to the vehicle inspection and maintenance location for a short period of time to avoid failure during the vehicle's journey.

[0056] When the functional evaluation value FA=0, the evaluation function is faulty, indicating that the device is completely unable to work and has lost its corresponding function. It must be repaired; otherwise, the vehicle cannot be started or driven.

[0057] The fault strategy module and maintenance strategy module mainly execute corresponding actions based on the functional evaluation value. If the functional evaluation value is 0.8 ≤ FA ≤ 1, the functional evaluation is normal, and no corresponding operation is performed. If the functional evaluation value is 0.2 < FA < 0.8, an abnormal functional evaluation is indicated, and the system will provide maintenance strategies based on the location of the abnormality in the corresponding device. The user can then perform the operation as needed. For example, although components such as oil filters and air filters may have wear and tear, they can still function normally without affecting their role. The system can provide solutions after searching the knowledge graph, such as cleaning or replacing the filter element. The user can perform maintenance according to actual needs. If the functional evaluation value is 0 < FA ≤ 0.2, a functional evaluation fault warning is issued, and the system will provide maintenance strategies based on the location of the abnormality in the corresponding device. If the user is unable to perform maintenance immediately, and if repair or maintenance is not possible, the vehicle can be driven briefly to a repair shop to prevent malfunctions during operation. For example, if components such as the oil filter or air filter are severely worn, immediate maintenance is required. The system can provide solutions after searching its knowledge graph, such as cleaning or replacing the filter. If the functional evaluation value FA=0, a functional fault is assessed. The system will provide repair strategies based on the fault status of the corresponding device, and the user must repair the affected device immediately. Otherwise, the vehicle will not start or move. For example, if components such as the oil filter or air filter are completely damaged, immediate repair is required. The system can provide solutions after searching its knowledge graph, such as replacing the filter. After the repair or maintenance of the affected device is completed, the system will re-monitor its operating status and begin a new cycle.

[0058] The above formulas are all numerical calculations after removing dimensions. The formulas are the closest to the real situation obtained by software simulation based on a large amount of data. The preset parameters and preset thresholds in the formulas are set by those skilled in the art according to the actual situation or obtained by simulation based on a large amount of data.

[0059] Example:

[0060] The recommended maintenance mileage for a certain type of vehicle is 5000 km. During normal driving, if the vehicle shows no abnormalities, no maintenance or repair will be performed. However, if a major malfunction occurs at this time, it could lead to serious consequences. This invention has been applied and verified in multiple vehicles of this type. After a long period of data collection and temporary handling, the following two typical handling cases are identified:

[0061] (1) When a vehicle reached 4250 km, the system, after data collection, extraction, classification, graph search, and correlation calculation, displayed a functional evaluation value (FA) of 0.6, indicating an oil filter malfunction and indicating oil filter blockage. The suggested maintenance strategy was to inspect, clean, and maintain the oil filter. After the vehicle stopped, relevant personnel inspected the oil filter and found it blocked by large impurities, causing increased oil filter pressure and poor filtration. After maintenance, the system restored the filter to normal without any abnormality warnings, and it could be used normally. If this vehicle did not have this system, the blocked oil filter would reduce oil flow, worsen filtration, cause increased engine wear and excessive engine temperature, affect engine lifespan, and in severe cases, cause engine failure, resulting in significant economic losses. This case fully demonstrates that the system can provide real-time alerts for vehicle malfunctions, promptly restore various vehicle performance parameters, extend vehicle lifespan, and maximize vehicle performance.

[0062] (2) When a vehicle reached 23,460 km, the system, after data collection, extraction, classification, graph search, and correlation calculation, displayed a functional evaluation value (FA) of 0.1, indicating an abnormal air filter with a high probability of damage. The recommended maintenance strategy was to immediately stop the vehicle and inspect and maintain the air filter. The relevant personnel immediately stopped the vehicle for inspection and maintenance, discovering that the air filter element was damaged, which would reduce the drag coefficient and worsen the air filtration effect. The filter element was then replaced without any further abnormalities, and the vehicle could be used normally. If this vehicle did not have this system, a damaged air filter element would increase intake impurities, which would enter the cylinders, causing wear on the engine cylinders and pistons, and in severe cases, engine seizure, resulting in engine damage and significant economic losses. This case fully demonstrates that the system can provide real-time alarms for temporary vehicle malfunctions, helping to promptly repair the vehicle, avoid significant losses, and potentially prevent various accidents.

[0063] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the invention by those skilled in the art. Any modifications, equivalent substitutions, or improvements made to the present invention should be included within the scope of protection of the present invention.

Claims

1. A vehicle health management system based on knowledge graph technology, characterized in that: The system includes a knowledge graph model, a data processing module, an intelligent search module, a health assessment module, a fault strategy module, and a maintenance strategy module. The knowledge graph model is used to establish a data model library for intelligent vehicle assessment and diagnosis. Through internally designed logical calculations, it aligns data with the status of various devices on the vehicle. Finally, it outputs the status of each device through intelligent retrieval, and can update the data in real time while simultaneously pushing out handling strategies. The data processing module is used to monitor the functions of various vehicle devices and process the data, providing instructions for executing the next action. The intelligent search module is used to match the attribute information and functional data of each device with the knowledge graph model, achieving data-physical alignment and accurately locating the functional position of devices with abnormal data. The health assessment module is used to determine whether each device can continue to work normally, assess the current working status of the device, and propose usage strategies; the fault strategy module and maintenance strategy module propose repair or maintenance strategies based on the functional assessment values ​​obtained by the health assessment module, and execute corresponding actions according to instructions.

2. The vehicle health management system based on knowledge graph technology according to claim 1, characterized in that: The knowledge graph model includes three sub-models: knowledge input, knowledge fusion, and knowledge output. In the knowledge input sub-model, two data model libraries are established: the original attributes of each device and the functional information of each device. First, based on the original attribute parameters of each device, an original attribute database of each device in the vehicle is established. Then, the working status data of each device in the vehicle under different usage stages and different operating conditions are monitored to establish a functional information database of each device under all usage stages and all operating conditions. The data in this database is divided into three data types: structured data, semi-structured data, and unstructured data.

3. The vehicle health management system based on knowledge graph technology according to claim 2, characterized in that: In the knowledge fusion sub-model, data is aligned with entities through logical calculations, and an entity data model is constructed. The health status of the device is inferred based on the data information and updated in real time. The data integrating the original attribute database and functional information database of the device, along with the semi-structured and unstructured entity-annotated data, are logically calculated to ensure that each data knowledge point corresponds one-to-one with the relevant features of each device entity on the vehicle. That is, different data combinations can accurately identify each device and its working status. Then, the functional data information is labeled, and different functional information data are assigned functional values ​​to achieve the quality assessment of each device.

4. The vehicle health management system based on knowledge graph technology according to claim 2, characterized in that: In the knowledge output sub-model, the data network from the knowledge fusion stage is interwoven into a knowledge graph, and knowledge is output through graph retrieval.

5. The vehicle health management system based on knowledge graph technology according to claim 1, characterized in that: The data processing module comprises three sub-modules: system monitoring, data acquisition, and data comparison. The system monitoring sub-module acquires attribute information of each device on the vehicle and monitors its functional status. Its key features are full vehicle coverage, full information coverage, and full functional coverage, providing a logical foundation for subsequent command actions. The data acquisition sub-module collects raw attribute information and functional status data of each device through data sensors deployed throughout the vehicle. Its key features are the speed, accuracy, and comprehensiveness of raw attribute information and functional status data acquisition, providing a data foundation for subsequent model calculations. The data comparison sub-module compares and extracts data with the original attribute information and functional data of each device through data visualization processing or simulation calculations to obtain information models of each device. Based on this, it calculates the corresponding functional anomaly curve for each device, determines whether each device is functionally abnormal based on this functional anomaly curve, and further executes subsequent logical actions. A functional anomaly identification model is established based on the information models of each device, and its expression is as follows: ； In the formula: n is the input data, which is the functional monitoring data of each device on the vehicle; FB(n) is the abnormal value of the function; When the output value FB(n) is 1, it is determined that the detection device is functioning normally and no corresponding operation is performed. When the output value FB(n) is 0, it is determined that the detected device is malfunctioning, and the system will continue to execute logical actions in the intelligent search module.

6. The vehicle health management system based on knowledge graph technology according to claim 1, characterized in that: The intelligent search module includes three sub-modules: data extraction, data classification, and graph search. The data extraction sub-module extracts abnormal functional data through logical calculation and data comparison, and locates the abnormal data to each device through attribute information, so as to achieve accurate alignment between data and physical objects. The data classification submodule categorizes device anomaly data into direct anomaly data and verification anomaly data. Direct anomaly data is directly aligned with device faults, indicating a functional malfunction. Verification anomaly data requires verification of the device's functionality, with verification results including both equipment-related and non-equipment-related causes, further determining whether the device is functionally malfunctioning. The knowledge graph search submodule searches, matches, and retrieves relevant device information data within the established vehicle knowledge graph model, accurately locating the functional anomalies of corresponding devices, providing precise information support for the next step of device health status assessment.

7. The vehicle health management system based on knowledge graph technology according to claim 1, characterized in that: The health assessment module includes two sub-modules: correlation calculation and anomaly evaluation. The correlation calculation sub-module performs correlation calculations on the abnormal data collected from the abnormal device and the corresponding data in the knowledge graph to further determine whether the device is abnormal and whether it can continue to work normally. The anomaly evaluation sub-module further evaluates the abnormal data determined by the correlation calculation sub-module to assess the health status of the device in order to take the next step.

8. The vehicle health management system based on knowledge graph technology according to claim 7, characterized in that: The aforementioned health assessment module first fits data from various devices on the vehicle to obtain a function curve function, denoted as GY(t), where t is time. Based on this function, a functional assessment model is established, the expression of which is: ； In the formula: FA is the functional evaluation value; δt is the coefficient function, and the coefficient function... t0 is the starting time, which is the time corresponding to the nearest t0 to the current time.

Images