An electrical system health state monitoring method, cloud server, storage medium and program product

Abstract

This invention discloses a method for monitoring the health status of an electrical system, a cloud server, a storage medium, and a program product. Based on multi-dimensional basic data, it identifies the actual operating conditions of a vehicle and extracts multi-dimensional actual electrical features such as parameter deviation characteristics, coordination characteristics, and spectral timing characteristics, achieving a quantitative description of the electrical system's health status. It introduces multi-dimensional benchmark electrical features of the same batch and model of the target electric vehicle under the same operating conditions as a statistical reference system, and determines anomalies by comparing actual features with the group statistical benchmark. This application's embodiments elevate electrical system health status identification from traditional isolated judgment of a single vehicle and fixed threshold comparison to a scientific dimension that integrates collective intelligence and adaptive operating conditions, effectively eliminating the influence of individual differences and changes in operating conditions on the judgment results, and significantly improving the accuracy and reliability of electrical anomaly identification.

Description

Technical Field

[0001] This invention relates to the field of electric vehicle technology, and in particular to a method for monitoring the health status of an electrical system, a cloud server, a storage medium, and a program product. Background Technology

[0002] Currently, health monitoring of electric vehicle electrical systems primarily relies on limited parameters (such as voltage and current) collected by onboard terminals, which are then compared with preset fixed thresholds to trigger fault alarms. However, electrical systems involve multiple interconnected modules, including high-voltage power, low-voltage control, and charging management. Monitoring single parameters and static thresholds is insufficient to capture coordinated anomalies between modules, nor can it distinguish the normal impact of changes in operating conditions (such as startup, driving, and charging) on ​​parameter fluctuations. Therefore, improving the accuracy of health status identification for electric vehicle electrical systems has become an urgent technical problem to be solved. Summary of the Invention

[0003] This application provides an electrical system health status monitoring method, cloud server, storage medium, and program product, which solves the technical problem in the prior art that the lack of a group statistical benchmark based on the same batch and model of vehicle and the ability to adapt to operating conditions leads to low accuracy in electrical system health status identification. It achieves the technical effect of significantly improving the accuracy of electrical system health status identification by constructing a group statistical feature baseline that matches the actual operating conditions.

[0004] In a first aspect, this application provides a method for monitoring the health status of an electrical system, applied to a cloud server, the method comprising: Acquire multi-dimensional basic data of the target electric vehicle in the current period; the multi-dimensional basic data includes high-voltage system data, low-voltage system data, charging system data, and preset correlation parameters; The actual operating conditions of the target electric vehicle are determined based on the aforementioned multi-dimensional basic data. Based on the aforementioned multi-dimensional basic data, the multi-dimensional actual electrical characteristics of the target electric vehicle are determined; the multi-dimensional actual electrical characteristics include parameter deviation characteristics, coordination characteristics, and spectral time-series characteristics. Based on the multi-dimensional actual electrical characteristics and the multi-dimensional benchmark electrical characteristics of the same batch and model of the target electric vehicle in the operating conditions that match the actual operating conditions, it is determined whether the target electric vehicle has an electrical abnormality in the current period. The multi-dimensional benchmark electrical characteristics are statistical characteristics of electrical characteristics in different dimensions.

[0005] Secondly, this application provides a cloud server, including: processor; Memory used to store the processor's executable instructions; The processor is configured to execute an electrical system health status monitoring method as provided in the first aspect.

[0006] Thirdly, this application provides a non-transitory computer-readable storage medium that, when the instructions in the storage medium are executed by the processor of a cloud server, enables the cloud server to execute an electrical system health status monitoring method as provided in the first aspect.

[0007] Fourthly, this application provides a computer program product, including computer instructions that are executed by a processor to implement an electrical system health status monitoring method as provided in the first aspect.

[0008] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: This application embodiment comprehensively captures the operational status information of each key module of the electric vehicle's electrical system by acquiring multi-dimensional basic data covering the high-voltage system, low-voltage system, charging system, and preset associated parameters, laying a complete data foundation for health monitoring. Furthermore, based on the multi-dimensional basic data, the actual operating conditions of the vehicle are identified, and multi-dimensional actual electrical features such as parameter deviation characteristics, coordination characteristics, and spectral time-series characteristics are extracted, achieving a quantitative description of the electrical system's health status. On this basis, multi-dimensional benchmark electrical features of the same batch and model of the target electric vehicle under the same operating conditions are introduced as a statistical reference system. Anomaly determination is made by comparing the actual features with the group statistical benchmark. Thus, this application embodiment elevates the identification of the electrical system's health status from traditional isolated judgment of a single vehicle and fixed threshold comparison to a scientific dimension that integrates collective wisdom and adaptive operating conditions, effectively eliminating the influence of individual differences and changes in operating conditions on the judgment results, and significantly improving the accuracy and reliability of electrical anomaly identification. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 A flowchart illustrating a method for monitoring the health status of an electrical system provided in this application embodiment; Figure 2 This is a schematic diagram of the structure of a cloud server provided in an embodiment of this application. Detailed Implementation

[0011] This application provides an electrical system health status monitoring method, cloud server, storage medium, and program product, which solves the technical problem in the prior art that the lack of a group statistical benchmark based on the same batch and model of vehicle and the ability to adapt to operating conditions leads to low accuracy in electrical system health status identification. It achieves the technical effect of significantly improving the accuracy of electrical system health status identification by constructing a group statistical feature baseline that matches the actual operating conditions.

[0012] The technical solution of this application embodiment is to solve the above-mentioned technical problems, and the general idea is as follows: This application embodiment comprehensively captures the operational status information of each key module of the electric vehicle's electrical system by acquiring multi-dimensional basic data covering the high-voltage system, low-voltage system, charging system, and preset associated parameters, laying a complete data foundation for health monitoring. Furthermore, based on the multi-dimensional basic data, the actual operating conditions of the vehicle are identified, and multi-dimensional actual electrical features such as parameter deviation characteristics, coordination characteristics, and spectral time-series characteristics are extracted, achieving a quantitative description of the electrical system's health status. On this basis, multi-dimensional benchmark electrical features of the same batch and model of the target electric vehicle under the same operating conditions are introduced as a statistical reference system. Anomaly determination is made by comparing the actual features with the group statistical benchmark. Thus, this application embodiment elevates the identification of the electrical system's health status from traditional isolated judgment of a single vehicle and fixed threshold comparison to a scientific dimension that integrates collective wisdom and adaptive operating conditions, effectively eliminating the influence of individual differences and changes in operating conditions on the judgment results, and significantly improving the accuracy and reliability of electrical anomaly identification.

[0013] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0014] First, it should be clarified that the term "and / or" in this article 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 article generally indicates that the preceding and following related objects have an "or" relationship.

[0015] This application provides a method for monitoring the health status of an electrical system, applied to a cloud server. The method includes steps S11-S14, which can be found in detail below. Figure 1 As shown.

[0016] Step S11: Obtain multi-dimensional basic data of the target electric vehicle in the current period; the multi-dimensional basic data includes high-voltage system data, low-voltage system data, charging system data, and preset correlation parameters; Step S12: Determine the actual operating conditions of the target electric vehicle based on the multi-dimensional basic data; Step S13: Determine the multi-dimensional actual electrical characteristics of the target electric vehicle based on the multi-dimensional basic data; the multi-dimensional actual electrical characteristics include parameter deviation characteristics, coordination characteristics, and spectral time series characteristics. Step S14: Based on the multi-dimensional actual electrical characteristics and the multi-dimensional benchmark electrical characteristics of the same model of the target electric vehicle in the same batch, corresponding to the operating conditions matching the actual operating conditions, determine whether the target electric vehicle has an electrical abnormality in the current period. The multi-dimensional benchmark electrical characteristics are statistical characteristics of electrical characteristics in different dimensions.

[0017] This application provides an electrical system health status monitoring method applied to a cloud server. The cloud server can communicate with multiple electric vehicles, which can be from different batches or the same batch, different models or the same model. By deploying a server in the cloud, it can aggregate massive amounts of electric vehicle operating data, constructing a group statistical benchmark (i.e., the multi-dimensional benchmark electrical characteristics involved in step S14) for each target electric vehicle based on the same batch and model, thus overcoming the limitations of local computing power and storage capacity of a single vehicle. The cloud server can dynamically identify the actual operating conditions of the vehicle based on the multi-dimensional basic data uploaded by the vehicle, and retrieve the group benchmark matching the operating conditions for comparison and analysis, thereby accurately distinguishing whether parameter fluctuations originate from individual aging, occasional disturbances, or changes in operating conditions. Furthermore, the cloud server can also utilize a cross-vehicle fault case library for feature matching and iteratively optimize the monitoring model based on feedback from the full dataset. As can be seen, the electrical system health status monitoring method provided in this application, which is executed by a cloud server, can improve the identification of the electrical system health status from isolated judgment of a single vehicle to accurate perception driven by collective intelligence, significantly improving the accuracy and generalization ability of identifying complex electrical anomalies (especially latent and collaborative faults).

[0018] Regarding step S11, obtain multi-dimensional basic data of the target electric vehicle in the current period; the multi-dimensional basic data includes high-voltage system data, low-voltage system data, charging system data, and preset correlation parameters.

[0019] The current period refers to the smallest time unit for the cloud server to monitor the electrical health status of the target electric vehicle. For example, it can be set to a period of 2 seconds to balance the real-time performance of data processing and system load. The cloud server acquires multi-dimensional basic data of the target electric vehicle in real time within the current period through a Telematics Box (T-Box). This data comprehensively covers all key modules of the electrical system.

[0020] Specifically, the high-voltage system data may include the bus voltage (U_high) of the power battery, the drive motor circuit current (I_motor), the battery circuit current (I_batt), the charging circuit current (I_charge), the insulation resistance of the high-voltage system (R_ins), the on / off state of the high-voltage contactor, and the temperature (T_power) of the insulated gate bipolar transistor (IGBT) of the power module, which are used to reflect the transmission and safety status of high-voltage energy.

[0021] The low-voltage system data includes the 12V battery voltage (U_low), the Controller Area Network (CAN) bus communication status (such as bus load rate and error frame count), the operating current of each Electronic Control Unit (ECU) (I_ecu1 to I_ecu8), and the on / off status of fuses, which are used to characterize the power supply and communication health of the low-voltage control network.

[0022] The charging system data includes the physical connection status of the charging gun, the charging connection confirmation signal (ControlPilot / Connection Confirmation, CP / CC), the real-time charging power change curve over time (P_charge-t), and the interaction messages between the vehicle and the charging pile, used to monitor the electrical safety and protocol compliance of the charging process. The preset associated parameters include the diagnostic fault codes (DTCs) reported by each controller, the current operating status of the vehicle (e.g., starting, driving, charging, stationary), the ambient temperature (T_amb), and the cumulative mileage, providing necessary contextual information for subsequent operating condition identification and feature analysis. Through the above process, the cloud server can completely capture the raw operating data of the target electric vehicle's electrical system in each monitoring cycle, including high-voltage, low-voltage, charging, and associated dimensions.

[0023] After acquiring the multi-dimensional basic data of the target electric vehicle for the current period, it needs to be preprocessed to remove outliers and ensure the accuracy and reliability of subsequent feature calculations. Specifically, firstly, data null values ​​or constant values ​​caused by momentary interruptions in sensor communication are filtered out to reduce the probability of misjudging communication anomalies as electrical faults; secondly, data with parameter jumps exceeding physical limits are removed, such as voltage values ​​exceeding the theoretical maximum value of battery cells connected in series, which usually originates from occasional failures of the acquisition unit; finally, logically contradictory records are removed, such as when the vehicle status is marked as "charging" while "driving signal activation" is detected, such conflicting data cannot reflect the true physical state. After the above cleaning, the retained high-quality multi-dimensional basic data will be used for subsequent operating condition partitioning and feature calculations.

[0024] Regarding step S12, the actual operating conditions of the target electric vehicle are determined based on the multi-dimensional basic data.

[0025] Based on cleaned, multi-dimensional basic data, the cloud server identifies the actual operating conditions of the target electric vehicle by analyzing the vehicle's status signals. When the key ON signal is detected but the READY signal is not yet activated, and the duration is less than 60 seconds, the vehicle is determined to be in the startup phase. During this phase, the electrical system is in a self-check and pre-charging process, and parameters fluctuate significantly. When the READY signal is activated and the vehicle speed is greater than 0, the vehicle is determined to be in the driving phase. At this time, the high-voltage system continuously outputs power, and the electrical load changes dynamically. When the charging gun is physically connected and the charging enable signal is activated, the vehicle is determined to be in the charging phase. During this phase, the battery management system interacts with the charging pile at high frequency. When the vehicle is turned off (key OFF), not charging, and this state lasts for more than 30 minutes, it is determined to be in the idle phase. At this time, the vehicle enters a low-power mode to monitor static leakage or battery self-discharge. This classification of operating conditions provides a foundation for matching targeted baseline electrical characteristics and threshold matrices for each subsequent phase.

[0026] Regarding step S13, the multi-dimensional actual electrical characteristics of the target electric vehicle are determined based on the multi-dimensional basic data; the multi-dimensional actual electrical characteristics include parameter deviation characteristics, coordination characteristics, and spectral time series characteristics.

[0027] Based on cleaned, multi-dimensional fundamental data, cloud servers extract multi-dimensional actual electrical characteristics that can quantify the health status of electrical systems through mathematical transformations and statistical analysis. These characteristics characterize the operating state of electrical systems from different dimensions, specifically including parameter deviation characteristics, coordination characteristics, and spectral time-series characteristics.

[0028] Parameter deviation characteristics are used to measure the degree of deviation of current electrical parameters from reference values, reflecting the aging or abnormal trends of the electrical system. For example, the high-voltage parameter deviation rate is the ratio of the actual bus voltage U_high to the reference voltage U_base under the same operating conditions, and the normal range is usually between 0.95 and 1.05; the low-voltage fluctuation index characterizes voltage stability by calculating the standard deviation σ_low of the 12V battery voltage over 30 seconds, and the normal value should be less than 0.3V; the insulation resistance decay rate is the ratio of the current insulation resistance R_ins to the initial value R_initial, and should be greater than 0.8 under healthy conditions.

[0029] Coordination characteristics are used to assess the consistency between different electrical modules or signals, revealing potential cross-system coupling faults. For example, power balance is the absolute value of the deviation rate between the high-voltage side output power P_high and the low-voltage side power consumption P_low, which should normally be less than 5%; signal synchronicity refers to the time difference Δt_sync between the accelerator pedal signal and the actual torque response of the motor, which should be less than 100ms under healthy conditions; module response consistency is assessed by measuring the standard deviation of the response time of each ECU under the same command, which should normally be less than 50ms.

[0030] Spectrum timing characteristics, on the other hand, detect latent faults or intermittent anomalies by analyzing the variation patterns of signals in the frequency and time domains. For example, the current harmonic distortion rate (THD) represents the proportion of harmonic components in the motor current, and should normally be less than 5%; the communication packet loss rate is measured by the number of error frames generated per second on the CAN bus, and should normally be less than 1 frame / second; the abnormal fluctuation frequency is a statistical measure of the number of times each electrical parameter exceeds the 3σ reference range per unit hour, and should normally be less than 2 times.

[0031] In step S13, the cloud server combines the multi-dimensional basic data of the current period and calculates the actual electrical characteristic values ​​of each item according to the above definition. Finally, it generates a feature vector containing multiple indicators in three major categories: parameter deviation, coordination, and spectrum timing, which is used for subsequent anomaly determination.

[0032] Furthermore, after completing the calculation of multi-dimensional actual electrical characteristics, the cloud server establishes an index according to the composite structure of "vehicle VIN, operating condition stage, and anomaly type" and stores the various characteristic parameters in an orderly manner to the cloud platform.

[0033] Specifically, the Vehicle Identification Number (VIN) is used as the primary index to ensure that the data is uniquely bound to a specific vehicle. The actual operating conditions identified in the current cycle (such as the start-up phase, driving phase, charging phase, or stationary phase) are used as the secondary index to facilitate subsequent retrieval of the corresponding baseline features for comparison according to the operating conditions. Finally, the abnormality type initially determined in step S13 (if no abnormality has been determined, it is marked as "normal") is used as the tertiary index to form a hierarchical storage key value.

[0034] Based on this index structure, the cloud server encapsulates all the feature parameters calculated in step S13 (such as high-voltage parameter deviation rate, low-voltage fluctuation index, power balance, current harmonic distortion rate, etc.) into feature vectors and persists them. Simultaneously, during storage, it automatically links to a fault case library for the same vehicle model. That is, by matching the current vehicle's VIN with the model information, it establishes a metadata link between the current feature vector and historical fault cases, providing an efficient data retrieval and pattern comparison foundation for subsequent deep matching based on the fault case library and model iteration based on maintenance feedback. Through this storage mechanism, the cloud platform constructs a structured electrical feature database based on vehicle, operating condition, and anomaly type, enabling refined management and rapid retrieval of massive amounts of vehicle data. The database can also contain the following: Electrical parameter baseline for the same model: Stores the parameter benchmark range (μ±3σ) under various operating conditions based on historical data statistics of the same batch of the same model, such as the high voltage parameter deviation rate of 0.95-1.05 and the low voltage fluctuation index <0.3V during the driving stage, as a group reference system for assessing the health status of individual vehicles.

[0035] Repair history: Records every repair event of the target electric vehicle from the time it leaves the factory to the present, including repair time, repair work order number, actual cause of failure, details of replaced parts, description of repair measures, etc., to provide historical context for subsequent failure case matching and root cause prediction.

[0036] Component replacement cycle: Stores the design life, historical replacement records, and average replacement mileage / time based on the same vehicle model for various electrical components (such as DC / DC converters, IGBT modules, 12V batteries, high-voltage relays, etc.) to assess the current aging degree and remaining life of the components.

[0037] Original monitoring data sequence: Stores multi-dimensional basic data (high voltage system data, low voltage system data, charging system data, and preset related parameters) uploaded by the target electric vehicle for each monitoring cycle in chronological order, forming a complete time-series data trajectory, supporting historical data backtracking and trend analysis.

[0038] Feature parameter vector library: Stores multi-dimensional actual electrical features (parameter deviation features, coordination features, and spectrum time series features) calculated for each monitoring cycle, and organizes them according to a composite index of "vehicle VIN, operating condition stage, and anomaly type" for easy retrieval and comparison.

[0039] Anomaly detection and early warning recording: Record the results of each anomaly detection, including the detection time, the type of preset condition that triggered it, the confirmed anomaly type, the corresponding early warning level, the response measures pushed, and the user / after-sales feedback status.

[0040] Electrical Fault Case Library: Stores all confirmed fault cases that have occurred in the same batch and model of vehicles. Each case includes the feature vector at the time of the fault, fault type label (such as high voltage leakage, communication interruption, power device aging, charging protocol abnormality), actual fault cause, repair measures, and post-repair tracking data, providing a reference template for feature matching degree calculation and root cause prediction.

[0041] Through the above structured design, the database achieves multi-dimensional data fusion from raw data to feature parameters, from individual trajectories to group statistics, and from historical cases to real-time judgments, providing relatively complete data support for every link of electrical system health status monitoring (condition identification, feature calculation, anomaly judgment, fault classification, health index assessment, and model iteration).

[0042] Regarding step S14, based on the multi-dimensional actual electrical characteristics and the multi-dimensional benchmark electrical characteristics of the same model of the target electric vehicle in the same batch, corresponding to the operating conditions matching the actual operating conditions, it is determined whether the target electric vehicle has an electrical abnormality in the current cycle. The multi-dimensional benchmark electrical characteristics are statistical characteristics of electrical characteristics in different dimensions.

[0043] There is a reference relationship between the multi-dimensional benchmark electrical characteristics and the multi-dimensional actual electrical characteristics. The multi-dimensional benchmark electrical characteristics are a standardized reference system constructed through statistical modeling based on historical operating data of the same batch and model of the target electric vehicle; while the multi-dimensional actual electrical characteristics are real-time quantitative indicators of the target electric vehicle in the current period. The two are established through operating condition matching, and the degree of deviation between the actual characteristic value and the benchmark range is used as the core basis for anomaly judgment.

[0044] The multi-dimensional baseline electrical characteristics are formally represented as parameter baseline ranges established for different operating conditions (such as the start-up phase, driving phase, charging phase, and stationary phase). Specifically, the cloud server pre-computes historical data of the same batch of vehicles under various operating conditions, calculates the mean μ and standard deviation σ of each electrical parameter using a normal distribution model, and then determines the normal fluctuation range as μ ± 3σ, which covers 99.7% of healthy samples. For example, for different vehicle operating conditions, this application embodiment sets differentiated electrical characteristic parameter baseline ranges to adapt to the different operating characteristics of the electrical system under various conditions. During the start-up phase, due to the pre-charging of the high-voltage system and the successive awakening of the low-voltage load, the parameter fluctuations are relatively large. Therefore, the normal range of the high-voltage parameter deviation rate (U_high / U_base) is set to 0.9 to 1.1, the 30-second standard deviation (σ_low) of the 12V battery voltage should be less than 0.5V, and the current harmonic distortion rate (THD) should be less than 8%. Once the driving phase begins, the electrical system stabilizes, and the threshold values ​​tighten accordingly. The normal range for high-voltage parameter deviation is adjusted to 0.95 to 1.05, the low-voltage fluctuation index must be less than 0.3V, and the current harmonic distortion rate must be less than 5%. During the charging phase, due to the connection to the external power grid and the fine-tuning of the battery management system, the parameter characteristics fall between those of starting and driving. Therefore, the normal range for high-voltage parameter deviation is set to 0.92 to 1.08, the low-voltage fluctuation index is less than 0.4V, and the current harmonic distortion rate is less than 6%. Through the aforementioned dynamic threshold matrix for different operating conditions, this embodiment can more accurately identify abnormal fluctuations in electrical parameters at each stage, effectively reducing the probability of misjudgments caused by changes in operating conditions.

[0045] In step S14, the cloud server first retrieves multi-dimensional benchmark electrical features (such as the μ±3σ range of each parameter, R_initial, etc. under the same operating condition) from the cloud database based on the actual operating condition identified in the current cycle of the target electric vehicle. Then, the multi-dimensional actual electrical features calculated in step S13 are compared one by one with the corresponding benchmark range. If all actual features fall within the μ±3σ range, it is preliminarily determined that there is no electrical abnormality in the current cycle.

[0046] Based on the initial determination in step S14 that the actual electrical characteristics do not match the benchmark range, a dual verification mechanism of continuous verification and engineering threshold is further introduced to improve the accuracy of anomaly identification. The specific process is divided into two stages: First stage: If at least one of the actual electrical features in the multi-dimensional actual electrical features does not match the corresponding reference electrical feature in the multi-dimensional reference electrical features, and the feature type corresponding to the actual electrical feature does not match the historical electrical feature in the preset historical period with the corresponding reference electrical feature, then it is determined that the target electric vehicle has a suspected electrical anomaly in the current period; the preset historical period includes at least two historical periods most recent to the current period; The second stage involves determining whether the target electric vehicle has any electrical abnormalities in the current period based on the multi-dimensional actual electrical characteristics and the preset thresholds corresponding to each electrical characteristic.

[0047] Regarding the first stage, suspected anomaly detection: The cloud server first checks whether at least one feature value in the multi-dimensional actual electrical characteristics of the current period exceeds its corresponding benchmark range (e.g., μ±3σ). If so, it further traces back the historical electrical feature values ​​of this feature within a preset historical period (e.g., the most recent three consecutive monitoring periods) and compares them with the same benchmark range. If this feature consistently exceeds the benchmark range in multiple consecutive periods, it is determined that the target electric vehicle has a "suspected electrical anomaly" in the current period. This design aims to eliminate occasional transient disturbances or sensor noise, ensuring that the abnormal signal has temporal persistence.

[0048] In other words, if any actual characteristic continuously deviates from the benchmark range (e.g., exceeding the limit for at least two consecutive monitoring cycles), the subsequent anomaly confirmation process is triggered. By introducing a statistical benchmark based on group data of the same batch and model, this step achieves adaptive elimination of individual differences and operating condition fluctuations, elevating the identification of the electrical system's health status from fixed threshold judgment to a scientific dimension of group statistical comparison, thereby fundamentally improving the accuracy and generalization ability of anomaly identification.

[0049] For example, suppose the high-voltage parameter deviation rate (U_high / U_base) of a target electric vehicle during the driving phase is 1.06, 1.07, and 1.06 in three consecutive monitoring cycles (the current cycle and the two preceding cycles), while the reference range for this operating condition is 0.95-1.05 (μ±3σ). Because it continuously exceeds the reference range, the system determines it as "suspected anomaly".

[0050] Regarding the second stage, anomaly confirmation and judgment. For vehicles marked as "suspected," the cloud server then compares the multi-dimensional actual electrical characteristics of the current period with preset engineering thresholds. These preset thresholds are usually more stringent than the statistical benchmark range (e.g., the upper limit of the voltage deviation rate benchmark is 1.05, while the preset threshold may be set to 1.1), and often require multiple dimensions of characteristics to exceed the limit simultaneously (e.g., both voltage deviation rate and power balance exceed the limit) before an anomaly is finally confirmed.

[0051] Specifically, the second stage process involves determining that the target electric vehicle has an electrical anomaly in the current period if the multi-dimensional actual electrical characteristics satisfy at least one condition in the preset condition set. In other words, the cloud server uses the preset condition set to perform a composite judgment on suspected anomalies. Each condition is designed for a specific fault mode, and its preset threshold is comprehensively calibrated based on engineering experience, safety standards, and statistical analysis. The following explains each condition and provides a reasonable range for each preset threshold.

[0052] The preset condition set includes: Condition 1: At least one parameter deviation feature in the multi-dimensional actual electrical features does not match the corresponding first preset threshold, and at least one synergy feature in the multi-dimensional actual electrical features does not match the corresponding second preset threshold. Condition 2: The insulation resistance attenuation rate, which belongs to the parameter deviation feature among the multi-dimensional actual electrical features, is less than the third preset threshold, and the communication packet loss rate, which belongs to the spectrum timing feature among the multi-dimensional actual electrical features, is greater than the fourth preset threshold. Condition 3: The frequency of abnormal fluctuations belonging to the spectrum timing features in the multi-dimensional actual electrical features is greater than the fifth preset threshold, and the current harmonic distortion rate belonging to the spectrum timing features in the multi-dimensional actual electrical features is greater than the sixth preset threshold. Condition 4: The matching degree between the multi-dimensional actual electrical characteristics and the electrical fault case characteristics of the same batch and model of the target electric vehicle is greater than the seventh preset threshold.

[0053] Regarding condition one, the aim is to detect cross-module coupling anomalies, such as simultaneous exceedances in high-voltage parameter deviation and power balance, indicating potential aging of power devices or control malfunctions. The first preset threshold is used to determine the severity of parameter deviations. For example, the upper limit threshold for the high-voltage parameter deviation rate (U_high / U_base) is typically set to 1.05 to 1.1. This means that exceeding the statistical baseline (μ±3σ upper limit 1.05) but not reaching the danger threshold will not trigger a trigger, while exceeding 1.1 is considered a significant deviation. The second preset threshold is used to determine anomalies in synergy characteristics. For example, the power balance deviation rate is typically set to 5% to 10%, a range that integrates sensor accuracy and normal system fluctuations. Exceeding this range indicates a non-negligible power mismatch between the high-voltage and low-voltage sides.

[0054] Regarding condition two, a warning is specifically issued for the risk of high-voltage leakage, as decreased insulation performance is often accompanied by electromagnetic interference leading to communication anomalies. The third preset threshold corresponds to the insulation resistance attenuation rate, typically set at 0.5 to 0.6. This means that when the current insulation resistance is lower than 50%-60% of the initial value, a serious leakage risk is considered to exist. This value is calculated based on the national standard's minimum requirement for the absolute value of insulation resistance. The fourth preset threshold corresponds to the CAN bus communication packet loss rate. Under normal operating conditions, the packet loss rate should be less than 1 frame / second, while the threshold for triggering the warning is typically set at 5 to 10 frames / second, indicating that communication is significantly interfered with, which may affect the reliable transmission of control commands.

[0055] Regarding condition three, it is used to identify aging or latent faults in power devices. Such faults often manifest as frequent parameter fluctuations and waveform distortion. The fifth preset threshold is for abnormal fluctuation frequency, that is, the number of times the parameter exceeds the 3σ reference range per unit hour. Normally, it should be less than 2 times / hour, and the trigger threshold can be set to 5 to 10 times / hour. High frequency fluctuations indicate a decrease in system stability. The sixth preset threshold is for current harmonic distortion (THD). Under normal operating conditions, it should be less than 5%, while the trigger threshold is usually set to 10% to 15%. Exceeding this value means that there is serious nonlinear distortion in the motor or inverter, which may lead to overheating or reduced efficiency.

[0056] Regarding condition four, pattern matching is performed based on a historical fault case library, which is suitable for the rapid identification of known fault types. The seventh preset threshold corresponds to the matching degree, which is usually set to 70% to 90% (e.g., 80%). This value is obtained through machine learning training and aims to balance recall and accuracy. When the similarity between the current feature vector and historical cases exceeds this threshold, it can be considered that the vehicle has the same or highly similar electrical anomaly as the historical case.

[0057] Through the above-mentioned composite judgment conditions, the embodiments of this application achieve accurate identification of various electrical anomaly modes, covering typical scenarios such as cross-system coupling faults, leakage risks, and device aging, and ensuring the reliability and real-time performance of the early warning through engineering calibration of preset thresholds.

[0058] For example, suppose the high-voltage parameter deviation rate (U_high / U_base) of a target electric vehicle during the driving phase is 1.06, 1.07, and 1.06 in three consecutive monitoring cycles (the current cycle and the two preceding cycles), while the baseline range for this condition is 0.95-1.05 (μ±3σ). Because it consistently exceeds the baseline range, the system determines it as "suspected anomaly." Subsequently, the system calls preset thresholds for verification: the first preset threshold for this model is 1.1 (corresponding to the threshold for voltage deviation rate exceeding the standard alone), but the current value of 1.06 does not exceed 1.1; however, the system simultaneously detects that the power balance in the current cycle is 12%, exceeding the second preset threshold (10%), and this condition belongs to condition one in the preset condition set. Therefore, the system ultimately confirms that the vehicle has an electrical anomaly. Conversely, if only the voltage deviation rate remains between 1.06 and 1.07 but the power balance is normal, and no other combined conditions are triggered, the suspected status is maintained, and a high-level warning is not triggered temporarily. Through this two-layer judgment mechanism, the embodiments of this application effectively reduce the false alarm rate while ensuring sensitivity.

[0059] Furthermore, after determining that the target electric vehicle has an electrical anomaly in the current cycle, the method further includes: Based on the multi-dimensional actual electrical features and the preset thresholds corresponding to each electrical feature, a set of abnormal electrical features that do not match the preset thresholds corresponding to each electrical feature is determined from the multi-dimensional actual electrical features. Based on the set of abnormal electrical features and the preset first correlation between abnormal electrical features and electrical fault types, the predicted electrical fault type of the target electric vehicle in the current period is determined. The electrical fault types include at least high-voltage leakage, communication interruption, power device aging, and charging protocol anomaly.

[0060] After determining that the target electric vehicle has an electrical anomaly in the current cycle, the cloud server further performs a fault type identification step to locate the specific module or cause of the anomaly. This process consists of two stages: Phase 1: Constructing an Abnormal Electrical Feature Set. The cloud server iterates through the multi-dimensional actual electrical features involved in the judgment in step S14, filters out all feature items that do not match the corresponding preset thresholds, and combines these feature items to form an "abnormal electrical feature set". For example, if the judgment condition triggered in the current period is "insulation resistance attenuation rate less than 0.6 and communication packet loss rate greater than 5 frames / second", then this set will include the two abnormal features {insulation resistance attenuation rate too low, communication packet loss rate too high}.

[0061] The second stage: Mapping and predicting electrical fault types. The cloud server invokes a pre-defined "first association relationship between abnormal electrical features and electrical fault types," matching the aforementioned set of abnormal features with patterns in the fault type library to determine the predicted electrical fault type for the current period. This association relationship is based on pre-established mapping rules using a large amount of historical fault data and maintenance cases, with each fault type corresponding to one or more typical combinations of abnormal features.

[0062] Specifically, regarding high-voltage leakage faults, when the abnormal electrical characteristic set includes both "insulation resistance attenuation rate < 0.6" and "communication packet loss rate > 5 frames / second", the system maps it to a high-voltage leakage fault. This is because severe degradation of insulation performance is often accompanied by electromagnetic interference, leading to abnormal CAN bus communication; the simultaneous occurrence of these two factors is a typical characteristic of high-voltage leakage.

[0063] Regarding communication interruption type, when the abnormal electrical characteristic set mainly exhibits a combination of multiple items such as "signal synchronization Δt_sync > 100ms", "module response time standard deviation > 50ms", and "communication packet loss rate > 5 frames / second", and is not accompanied by obvious voltage or current deviation, the system maps it to a communication interruption type fault, pointing to a communication abnormality in the CAN network or a specific ECU.

[0064] Regarding power device aging type, when the abnormal electrical characteristic set contains multiple combinations of "high voltage parameter deviation rate > 1.1" or "< 0.9", "power balance > 10%", "current harmonic distortion rate THD > 15%" and "abnormal fluctuation frequency > 10 times / hour", the system maps it to a power device aging type fault, indicating that power components such as IGBTs, relays or motor controllers may have performance degradation.

[0065] Regarding abnormal charging protocols, when the set of abnormal electrical features contains features directly related to the charging system, such as when the charging gun is physically connected and charging is enabled, the charging power continuously deviates from the expected curve (abnormal P_charge-t), or the charging pile-vehicle interaction message is parsed to reveal a protocol incompatibility flag, and the condition of "matching degree with electrical fault case features of the same model > 80%" is met, the system maps it to an abnormal charging protocol fault.

[0066] Through the above two steps, the embodiments of this application transform the abstract set of abnormal features into fault types with clear maintenance directions, providing a decision-making basis for subsequent graded early warning and precise maintenance.

[0067] After determining the predicted electrical fault type of the target electric vehicle corresponding to the current period, the method further includes: Based on the predicted electrical fault type of the target electric vehicle in the current period and the second correlation between the preset electrical fault type and the electrical anomaly warning level measures, the target warning level measures corresponding to the target electric vehicle in the current period are determined. Control the target electric vehicle to operate in accordance with the target warning level measures.

[0068] After determining the predicted electrical fault type for the target electric vehicle in the current cycle, the cloud server further executes a tiered early warning and response measure determination step, mapping the fault type to a specific early warning level and corresponding control measures to achieve differentiated risk management. This process consists of two stages: Phase 1: Determining Target Early Warning Level Measures. Based on the predicted electrical fault type identified in the current period and the degree of exceedance of specific parameters associated with that fault, the cloud server invokes a pre-defined "second correlation between electrical fault type and electrical anomaly early warning level measures" to match the target early warning level measures for the current period. This correlation is a multi-dimensional decision matrix. Its inputs include the fault type (e.g., high-voltage leakage), the severity of specific exceedance characteristics (e.g., the specific value of insulation resistance), and whether it involves immediate safety risks (e.g., whether it may lead to electric shock or power interruption). The output is a predefined three-level early warning system level and its corresponding push targets and response measures.

[0069] Phase Two: Controlling the Vehicle to Execute Response Measures. The cloud server sends the determined warning level measures to the on-board terminal of the target electric vehicle through the vehicle-cloud communication link, and simultaneously pushes them to the user's mobile application (APP) and after-sales system. The vehicle executes the corresponding safety control strategies according to the instructions.

[0070] The three-tiered early warning system includes a Level 3 warning, a Level 2 warning, and a Level 1 warning, with Level 3 being the most severe, followed by Level 2, and Level 1 being the least severe. Specific examples are provided below to illustrate the scenarios for each level of early warning.

[0071] Example of a Level 3 Early Warning (High-Risk Immediate Response): If the current fault is determined to be a "high-voltage leakage type," and the insulation resistance R_ins is specifically monitored to be below 500Ω (trigger condition), this value has crossed the personal safety red line. The cloud server maps it to a Level 3 early warning through a second correlation. Subsequently, the cloud server issues a "Start Safety Mode Immediately" command to the vehicle terminal, the vehicle performs high-voltage power cut-off or restricts starting operations, and displays the fault code on the dashboard; simultaneously, the cloud server pushes the early warning information to the after-sales system, requiring after-sales personnel to respond and arrange emergency rescue within 1 hour. The core objective of this level is to ensure personal and vehicle safety.

[0072] Example of a Level 2 Early Warning (Medium-Risk Maintenance Handling): If the current cycle is determined to be a "power device aging" fault, but specifically manifests as a low-voltage fluctuation index σ_low = 0.6V (exceeding the normal value of 0.3V but not causing immediate risk), and there are no safety issues such as insulation degradation, the cloud server maps it to a Level 2 early warning. Subsequently, the cloud server pushes a notification to the user's APP stating "Electrical system abnormal, it is recommended to perform maintenance as soon as possible," reminding the user to arrange maintenance; at the same time, it automatically generates a maintenance work order for the after-sales system and prepares materials in advance according to the spare parts list associated with the fault type (such as DC / DC converters, 12V batteries, etc.). The core objective of this level is to optimize maintenance efficiency and user experience.

[0073] Example of Level 1 Warning (Low-Risk Monitoring and Handling): If the current period is only in the "suspected anomaly" stage, for example, a certain characteristic parameter exceeds the statistical baseline for multiple consecutive periods but has not yet reached the preset threshold, or a trend of exceeding the standard is identified (such as insulation resistance continuously and slowly decreasing but still above the safety line), the cloud server maps it to a Level 1 warning. This warning is only pushed to the cloud platform and the vehicle manufacturer's backend, without disturbing the user. The cloud platform then initiates enhanced monitoring of the vehicle, increasing the data sampling frequency from the normal period of 1Hz to 10Hz, and generating a trend analysis report; at the same time, the backend system accumulates the number of such warnings. When the same type of anomaly triggers 5 warnings in total, the remote upgrade (OTA, Over-The-Air) process of the control strategy is triggered to optimize the monitoring logic on the vehicle side. The core objective of this level is to achieve early detection of hidden dangers and iterative optimization of algorithms.

[0074] Through the aforementioned graded early warning and response mechanism, this application embodiment achieves a complete closed loop from fault type identification to differentiated handling: the third-level early warning focuses on the bottom line of safety, the second-level early warning optimizes maintenance services, and the first-level early warning supports algorithm evolution. While ensuring safety, it significantly improves the intelligence and refinement of electrical system health management.

[0075] Furthermore, after determining the multi-dimensional actual electrical characteristics of the target electric vehicle based on the aforementioned multi-dimensional basic data, the method further includes: Based on the multi-dimensional actual electrical characteristics and the historical abnormal data of electrical anomalies of the target electric vehicle within a preset historical time or preset historical mileage, the predicted health index of the electrical system of the target electric vehicle in the current period is determined by a multi-dimensional data weighting method. When the preset health index is less than the preset index threshold, the actual matching degree between the multi-dimensional actual electrical characteristics and the characteristics of different electrical fault cases of the same batch and model of the target electric vehicle is determined. From at least one electrical fault case with an actual matching degree greater than the preset matching degree, the electrical fault case with the largest actual matching degree is selected, and the electrical abnormality cause corresponding to the electrical fault case with the largest actual matching degree is determined as the predicted electrical abnormality cause of the target electric vehicle.

[0076] After determining the multi-dimensional actual electrical characteristics of the target electric vehicle, the cloud server further performs a health quantification assessment and deep diagnostic steps, transforming the multi-dimensional characteristics into an intuitive health score, and triggering root cause analysis based on a case library when the score is too low. This process consists of two phases: Phase 1: Predictive Health Index Calculation. The cloud server calculates the predicted health index of the target electric vehicle's electrical system for the current period based on multi-dimensional actual electrical characteristics (such as parameter deviation characteristics, coordination characteristics, and spectral time-series characteristics) obtained in the current period, combined with historical electrical anomaly data (such as historical anomaly frequency and historical warning levels) recorded within a preset historical time period or mileage (e.g., the past 30 days or a cumulative mileage of 5000 kilometers). This is achieved through a multi-dimensional data weighting method. The index uses a quantitative scoring format from 0 to 100 points, where 100 points represents a perfect state completely consistent with the benchmark of new vehicles in the same batch, and a lower score indicates a worse health condition. During the weighted calculation, the weight of each feature dimension is allocated according to its impact on the electrical system's lifespan and safety. For example, the weight of insulation resistance attenuation rate is usually higher than that of the low-voltage fluctuation index, and frequently occurring historical anomalies in the recent period will incur additional penalty coefficients on the score.

[0077] Phase Two: Deep Matching and Root Cause Prediction. When the predicted health index calculated above is lower than a preset threshold (e.g., 60 points), it indicates a significant potential risk in the vehicle's electrical system, requiring further clarification of the cause of the anomaly. The cloud server then initiates a deep diagnostic process: it iterates through and compares the multi-dimensional actual electrical feature vector of the current period with the features of different electrical fault cases of the same model in the same batch stored in the cloud, calculating the actual matching degree of each match (e.g., using Euclidean distance or cosine similarity algorithms). From all fault cases with an actual matching degree greater than a preset matching degree (e.g., 80%), the fault case with the highest actual matching degree is selected, and the electrical anomaly cause recorded in this case is determined as the predicted electrical anomaly cause for the current target electric vehicle.

[0078] For example, suppose a target electric vehicle's predicted health index is 58 points in the current period, lower than the preset threshold of 60 points, triggering deep diagnosis. The cloud server compares its current feature vector (including high-voltage parameter deviation rate of 1.08, power balance of 12%, THD of 14%, and abnormal fluctuation frequency of 8 times / hour, etc.) with 500 historical cases in the same model's fault case library. The calculation shows that the case numbered "DC-DC-2024-015" in the case library has a matching degree of 87% (exceeding the preset matching degree of 80%), which is the highest among all cases. The anomaly recorded in this historical case is "aging of the internal MOSFET of the DC / DC converter leading to unstable output." Based on this, the cloud server identifies "aging of the internal MOSFET of the DC / DC converter" as the predicted electrical anomaly cause of the target electric vehicle, providing precise guidance for subsequent maintenance work order generation and spare parts preparation. Through this process, this embodiment of the application transforms abstract electrical characteristics into root cause predictions with clear maintenance directions, achieving a deep intelligent upgrade from "discovering anomalies" to "locating the cause."

[0079] After determining the electrical anomaly cause corresponding to the electrical fault case with the highest actual matching degree as the predicted electrical anomaly cause of the target electric vehicle, the method further includes: Obtain the actual causes of failures in the target electric vehicle's subsequent repair work orders; The preset parameters in the multi-dimensional data weighting method are corrected based on the difference between the actual fault cause and the predicted electrical anomaly cause.

[0080] After identifying the electrical fault case with the highest actual matching degree as the predicted electrical fault cause for the target electric vehicle, the cloud server further executes a model iteration and optimization step. Through comparison and feedback between actual repair results and predicted results, the accuracy and adaptability of the diagnostic model are continuously improved. This process consists of two stages: Phase 1: Obtaining Actual Repair Feedback. Once the target electric vehicle enters the after-sales repair process and the fault is repaired, the repair personnel will fill out a repair work order, recording the actual cause of the fault, the replaced parts, and the repair measures. The cloud server obtains the actual fault cause from this repair work order through the after-sales system interface and uses it as the benchmark truth value for evaluating the accuracy of the prediction.

[0081] The second stage involves refining the model parameters. The cloud server compares the actual cause of the fault with the previously generated predicted causes of electrical anomalies, analyzing the differences between the two. If inconsistencies exist, it indicates a deviation in the current multi-dimensional data weighting method (i.e., the health index calculation model), requiring correction. Based on the difference analysis results, the cloud server automatically adjusts the preset parameters in the multi-dimensional data weighting method. This includes adjusting the weight coefficients of various electrical characteristics in the health index calculation, correcting the mapping rules between abnormal electrical characteristics and fault types, or optimizing the preset matching threshold. Through this closed-loop feedback mechanism, each maintenance provides the model with a learning opportunity, making the diagnosis of subsequent anomalies of the same type more accurate.

[0082] For example, suppose a target electric vehicle was previously predicted to have a "power device aging" fault, with the predicted cause of the electrical anomaly being "IGBT module performance degradation" (based on features such as THD > 15% and abnormal fluctuation frequency > 10 times / hour). However, after the vehicle was brought in for repair, the actual repair order recorded the fault as "poor contact of the CAN transceiver inside the motor controller" (belonging to the "communication interruption type"). After the cloud server obtained the actual fault cause, it found that it did not match the prediction result, triggering the model correction process. Analysis revealed that the current model assigned too much weight to the feature "high THD," leading to a misclassification as power device aging; while the correlation between "abnormal fluctuation frequency" and "communication packet loss rate" in the actual fault was not fully captured. Based on this, the cloud server automatically adjusts the weighting parameters: appropriately reducing the weight of THD features in the health index calculation, while increasing the collaborative weight of "communication packet loss rate" and "abnormal fluctuation frequency," and updating the mapping table between abnormal features and fault types. This makes it more likely to classify subsequent feature combinations such as "high communication packet loss rate + high abnormal fluctuation frequency" as "communication interruption type" rather than "power device aging type." Through this continuous iteration, the accuracy of the diagnostic model is constantly improved, ultimately achieving a high-performance indicator of "false negative rate <2%."

[0083] In summary, this application embodiment comprehensively captures the operational status information of each key module of the electric vehicle's electrical system by acquiring multi-dimensional basic data covering the high-voltage system, low-voltage system, charging system, and preset associated parameters, laying a complete data foundation for health monitoring. Furthermore, based on the multi-dimensional basic data, the actual operating conditions of the vehicle are identified, and multi-dimensional actual electrical features such as parameter deviation characteristics, coordination characteristics, and spectral time-series characteristics are extracted, achieving a quantitative description of the electrical system's health status. On this basis, multi-dimensional benchmark electrical features of the same batch and model of the target electric vehicle under the same operating conditions are introduced as a statistical reference system, and anomaly determination is made by comparing the actual features with the group statistical benchmark. Thus, this application embodiment elevates the identification of the electrical system's health status from traditional isolated judgment of a single vehicle and fixed threshold comparison to a scientific dimension that integrates collective wisdom and adaptive operating conditions, effectively eliminating the influence of individual differences and changes in operating conditions on the judgment results, and significantly improving the accuracy and reliability of electrical anomaly identification.

[0084] This application's embodiments achieve intelligent and precise monitoring and management of the health status of electric vehicle electrical systems by constructing a complete technical closed loop covering data acquisition, feature extraction, anomaly detection, fault classification, graded early warning, and model iteration. First, based on the acquisition and operating condition identification of multi-dimensional basic data, combined with the statistical benchmark of the same batch and model of vehicles, continuous verification and composite threshold judgment mechanisms significantly reduce the false alarm rate while ensuring sensitivity. Second, by constructing an abnormal electrical feature set and mapping it to preset fault types, precise location of specific fault modes such as high-voltage leakage, communication interruption, power device aging, and charging protocol anomalies is achieved, providing clear guidance for maintenance. Furthermore, based on a graded early warning system matching fault type and parameter severity, differentiated response measures such as high-voltage power cut-off, user reminders, and enhanced monitoring are implemented, optimizing user experience and maintenance efficiency while ensuring safety. In addition, by constructing a structured database containing electrical parameter baselines, maintenance records, component replacement cycles, and a fault case library, combined with quantitative assessment of health indices and deep case matching, predictive diagnosis of the root causes of anomalies is achieved. Ultimately, by using actual maintenance feedback to make closed-loop corrections to the weighted parameters, the diagnostic model is driven to continuously iterate and evolve, effectively controlling the system's false alarm rate and achieving a fundamental leap from "passive alarm" to "proactive early warning and continuous evolution".

[0085] The electric vehicle electrical system health status monitoring method provided in this application achieves significant technical effects through multi-dimensional feature fusion and intelligent analysis. In terms of early warning performance, multi-dimensional feature fusion increases the accuracy of electrical anomaly identification from 65% to 96% compared to traditional methods, with a false negative rate of less than 3%, and can detect potential faults 3 to 6 months in advance, achieving a leap from passive response to proactive early warning. Regarding maintenance support, this application embodiment can accurately locate anomalies to the module level, reducing the average troubleshooting time from 4 hours to 45 minutes and lowering maintenance costs by 65%, significantly improving after-sales maintenance efficiency. In terms of safety protection, the early warning response time for fatal faults such as high-voltage leakage is less than 200 milliseconds, reducing the risk of electric shock and fire by 90%, effectively protecting the safety of passengers. Regarding range maintenance, by proactively repairing power loss anomalies, the actual driving range retention rate is increased by 8% to 10%, directly improving the user experience. In terms of overall user experience, road breakdowns caused by electrical faults are reduced by 85%, and user satisfaction is increased by 40%. Furthermore, the fault big data accumulated in the embodiments of this application can be fed back into the design of new vehicles, reducing the electrical failure rate of new models by 35%; at the same time, the solution is compatible with pure electric and hybrid models, reducing the integration cost for car manufacturers by 40%.

[0086] Based on the same inventive concept, the embodiments of this application provide, as follows: Figure 2 The cloud server shown includes: Processor 21; Memory 22 is used to store executable instructions of the processor 21; The processor 21 is configured to execute an electrical system health status monitoring method as described above.

[0087] Based on the same inventive concept, embodiments of this application provide a non-transitory computer-readable storage medium that, when the instructions in the storage medium are executed by the processor of a cloud server, enables the cloud server to execute an electrical system health status monitoring method as described above.

[0088] Based on the same inventive concept, embodiments of this application provide a computer program product, including computer instructions, which are executed by a processor to implement an electrical system health status monitoring method as described above.

[0089] Since the cloud server described in this embodiment is the cloud server used to implement the information processing method in the embodiments of this application, those skilled in the art can understand the specific implementation method and various variations of the cloud server in this embodiment based on the information processing method described in the embodiments of this application. Therefore, how the cloud server implements the method in the embodiments of this application will not be described in detail here. Any cloud server used by those skilled in the art to implement the information processing method in the embodiments of this application falls within the scope of protection of this application.

[0090] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0091] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0092] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0093] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1The steps of the function specified in one or more boxes.

[0094] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the invention.

[0095] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for monitoring the health status of an electrical system, characterized in that, Applied to cloud servers, the method includes: Acquire multi-dimensional basic data of the target electric vehicle in the current period; the multi-dimensional basic data includes high-voltage system data, low-voltage system data, charging system data, and preset correlation parameters; The actual operating conditions of the target electric vehicle are determined based on the aforementioned multi-dimensional basic data. Based on the aforementioned multi-dimensional basic data, the multi-dimensional actual electrical characteristics of the target electric vehicle are determined; the multi-dimensional actual electrical characteristics include parameter deviation characteristics, coordination characteristics, and spectral time-series characteristics. Based on the multi-dimensional actual electrical characteristics and the multi-dimensional benchmark electrical characteristics of the same batch and model of the target electric vehicle in the operating conditions that match the actual operating conditions, it is determined whether the target electric vehicle has an electrical abnormality in the current period. The multi-dimensional benchmark electrical characteristics are statistical characteristics of electrical characteristics in different dimensions.

2. The method for monitoring the health status of an electrical system as described in claim 1, characterized in that, Based on the aforementioned multi-dimensional actual electrical characteristics and the multi-dimensional benchmark electrical characteristics of the same batch and model of the target electric vehicle corresponding to the operating conditions matching the actual operating conditions, it is determined whether the target electric vehicle has any electrical anomalies in the current period, including: If at least one of the multi-dimensional actual electrical features does not match the corresponding reference electrical feature in the multi-dimensional reference electrical features, and the historical electrical features of the feature type corresponding to the actual electrical feature in the preset historical period do not match the corresponding reference electrical feature, then it is determined that the target electric vehicle has a suspected electrical anomaly in the current period; the preset historical period includes at least two historical periods most recent to the current period; Based on the multi-dimensional actual electrical characteristics and the preset thresholds corresponding to each electrical characteristic, it is determined whether the target electric vehicle has an electrical anomaly in the current period.

3. The method for monitoring the health status of an electrical system as described in claim 2, characterized in that, Based on the multi-dimensional actual electrical characteristics and the preset thresholds corresponding to each electrical characteristic, it is determined whether the target electric vehicle has an electrical anomaly in the current period, including: If the multi-dimensional actual electrical characteristics satisfy at least one condition in the preset condition set, then it is determined that the target electric vehicle has an electrical anomaly in the current period; The preset condition set includes: At least one parameter deviation feature in the multi-dimensional actual electrical features does not match the corresponding first preset threshold, and at least one synergy feature in the multi-dimensional actual electrical features does not match the corresponding second preset threshold; The insulation resistance attenuation rate, which belongs to the parameter deviation feature among the multi-dimensional actual electrical features, is less than the third preset threshold, and the communication packet loss rate, which belongs to the spectrum timing feature among the multi-dimensional actual electrical features, is greater than the fourth preset threshold. The frequency of abnormal fluctuations belonging to the spectrum timing features in the multi-dimensional actual electrical features is greater than the fifth preset threshold, and the current harmonic distortion rate belonging to the spectrum timing features in the multi-dimensional actual electrical features is greater than the sixth preset threshold. The matching degree between the multi-dimensional actual electrical characteristics and the electrical fault case characteristics of the same batch and model of the target electric vehicle is greater than the seventh preset threshold.

4. A method for monitoring the health status of an electrical system as described in any one of claims 1-3, characterized in that, After determining that the target electric vehicle has an electrical anomaly in the current period, the method further includes: Based on the multi-dimensional actual electrical features and the preset thresholds corresponding to each electrical feature, a set of abnormal electrical features that do not match the preset thresholds corresponding to each electrical feature is determined from the multi-dimensional actual electrical features. Based on the set of abnormal electrical features and the first correlation between the preset abnormal electrical features and electrical fault types, the predicted electrical fault type of the target electric vehicle in the current period is determined.

5. The method for monitoring the health status of an electrical system as described in claim 4, characterized in that, After determining the predicted electrical fault type of the target electric vehicle corresponding to the current period, the method further includes: Based on the predicted electrical fault type of the target electric vehicle in the current period and the second correlation between the preset electrical fault type and the electrical anomaly warning level measures, the target warning level measures corresponding to the target electric vehicle in the current period are determined. Control the target electric vehicle to operate in accordance with the target warning level measures.

6. The method for monitoring the health status of an electrical system as described in claim 1, characterized in that, After determining the multi-dimensional actual electrical characteristics of the target electric vehicle based on the aforementioned multi-dimensional basic data, the method further includes: Based on the multi-dimensional actual electrical characteristics and the historical abnormal data of electrical anomalies of the target electric vehicle within a preset historical time or preset historical mileage, the predicted health index of the electrical system of the target electric vehicle in the current period is determined by a multi-dimensional data weighting method. When the preset health index is less than the preset index threshold, the actual matching degree between the multi-dimensional actual electrical characteristics and the characteristics of different electrical fault cases of the same batch and model of the target electric vehicle is determined. From at least one electrical fault case with an actual matching degree greater than the preset matching degree, the electrical fault case with the largest actual matching degree is selected, and the electrical abnormality cause corresponding to the electrical fault case with the largest actual matching degree is determined as the predicted electrical abnormality cause of the target electric vehicle.

7. The method for monitoring the health status of an electrical system as described in claim 6, characterized in that, After determining the electrical anomaly cause corresponding to the electrical fault case with the highest actual matching degree as the predicted electrical anomaly cause of the target electric vehicle, the method further includes: Obtain the actual causes of failures in the target electric vehicle's subsequent repair work orders; The preset parameters in the multi-dimensional data weighting method are corrected based on the difference between the actual fault cause and the predicted electrical anomaly cause.

8. A cloud server, characterized in that, include: processor; Memory used to store the processor's executable instructions; The processor is configured to execute an electrical system health status monitoring method as described in any one of claims 1 to 7.

9. A non-transitory computer-readable storage medium, characterized in that, When the instructions in the storage medium are executed by the processor of the cloud server, the cloud server is able to execute an electrical system health status monitoring method as described in any one of claims 1 to 7.

10. A computer program product, characterized in that, It includes computer instructions, which are executed by a processor to implement an electrical system health status monitoring method as described in any one of claims 1 to 7.

Images