Method and system for predicting power-on fault of vehicle and vehicle

Abstract

The invention discloses a vehicle power-on fault prediction method and system and a vehicle, and the method comprises the steps: collecting vehicle operation data and environment temperature data, the vehicle operation data comprising vehicle part data, battery data and user behavior data; based on the vehicle operation data and the environment temperature data, preset evaluation models corresponding to the vehicle part data, the battery data and the user behavior data are called for risk evaluation, a corresponding risk evaluation result is obtained, and the preset evaluation models comprise a battery evaluation model, a part evaluation model and a user behavior risk evaluation model; performing fusion analysis on the risk assessment result to generate a comprehensive risk assessment result; and predicting the risk level of the power-on fault of the vehicle according to the comprehensive risk assessment result. Potential risks can be recognized in advance before a fault occurs, conversion from passive maintenance to active early warning is achieved, vehicle using interruption is effectively avoided, and the user experience and the whole vehicle operation reliability are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of vehicle technology, and in particular to a method, system, and vehicle for predicting vehicle power-on faults. Background Technology

[0002] With the increasing popularity of electric vehicles, the inability to power on a vehicle—that is, the inability to start or operate normally due to abnormal power supply from the low-voltage system—has become a critical issue affecting user experience and overall vehicle reliability. The causes of this fault are complex, primarily involving multiple factors such as a depleted 12V lead-acid battery, failures in key components like relays, and improper electricity and vehicle usage by the user.

[0003] However, current technologies mostly address this issue through reactive, post-event responses, lacking effective pre-event prediction and early warning capabilities. While some solutions attempt to monitor and issue warnings by setting battery voltage thresholds, these methods rely on a single parameter, resulting in overly simplistic judgment logic and low accuracy and reliability of warnings. This makes it difficult to achieve early identification and effective intervention of potential risks. Summary of the Invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art.

[0005] Therefore, one objective of this invention is to propose a method for predicting vehicle power-on failures. This method can identify potential risks in advance before the failure occurs, realize the transformation from passive maintenance to proactive early warning, reserve sufficient response time for users, effectively avoid vehicle interruption, and significantly improve user experience and the reliability of vehicle operation.

[0006] Therefore, a second objective of the present invention is to provide a predictive system for vehicle power-on failures.

[0007] Therefore, a third objective of the present invention is to provide a vehicle.

[0008] To achieve the above objectives, a first aspect of the present invention discloses a method for predicting vehicle power-on failures, comprising: collecting vehicle operating data and ambient temperature data, wherein the vehicle operating data includes vehicle component data, battery data, and user behavior data; based on the vehicle operating data and the ambient temperature data, calling a preset assessment model corresponding to the vehicle component data, the battery data, and the user behavior data to perform a risk assessment, thereby obtaining a corresponding risk assessment result, wherein the preset assessment model includes a battery assessment model, a component assessment model, and a user behavior risk assessment model; performing a weighted fusion analysis on the risk assessment result to generate a comprehensive risk assessment result; and predicting the risk level of the vehicle power-on failure based on the comprehensive risk assessment result.

[0009] The vehicle power-on fault prediction method according to embodiments of the present invention, by collecting vehicle operation data and ambient temperature data, can comprehensively acquire key information reflecting the vehicle status and usage environment, effectively reducing the false alarm and false negative rates caused by traditional single-parameter judgment. Simultaneously, based on user behavior modeling, it performs refined analysis of different users' driving habits, achieving personalized risk assessment and providing accurate early warnings and customized suggestions, improving the targeting and effectiveness of management. Based on this, based on vehicle operation data and ambient temperature data, it calls preset assessment models corresponding to vehicle component data, battery data, and user behavior data respectively, conducting independent risk assessments from multiple dimensions to obtain risk assessment results for each dimension. Subsequently, the risk assessment results for each dimension are normalized and weighted fusion analysis to generate a comprehensive risk assessment result. Finally, based on this comprehensive risk assessment result, the risk level of a vehicle power-on fault is predicted. This allows for early identification of potential risks before a fault occurs, realizing a shift from passive maintenance to proactive early warning, providing users with sufficient response time, effectively avoiding vehicle interruption, and significantly improving user experience and overall vehicle reliability.

[0010] In addition, the vehicle power-on fault prediction method according to the above embodiments of the present invention may also have the following additional technical features: In some embodiments, the battery data includes the vehicle's resting voltage data during inactivity, the DC-DC (Direct Current to Direct Current Converter) output voltage, the DC-DC output current, and the vehicle's battery voltage. When performing a risk assessment based on the battery data and the ambient temperature data using the battery assessment model to obtain the corresponding risk assessment result, the process includes: assessing the battery's charge / discharge abnormality risk based on the DC-DC output voltage, the DC-DC output current, and the battery voltage; predicting the natural charge loss rate per unit time using a preset regression algorithm based on the resting voltage data and the ambient temperature data, and assessing the battery's self-discharge abnormality risk based on the natural charge loss rate; obtaining the battery's power-off time and the number of times recharging was triggered during the power-off period; establishing a mapping relationship between the recharging frequency and the battery's health status using a preset reliability distribution model based on the power-off time, the number of recharging times, and the ambient temperature data; and predicting the current battery health status value based on the mapping relationship to assess the battery's capacity. Battery degradation risk; Time series data is acquired, and a preset time series analysis algorithm is used to predict the voltage change trend of the battery. When the predicted battery voltage is lower than a preset power-on failure threshold, a start-up failure risk is assessed. The time series data includes historical battery voltage data, historical ambient temperature data, and historical user behavior data. The actual battery natural loss rate and the actual battery health status value are acquired. A first residual between the actual battery natural loss rate and the predicted battery natural loss rate are determined, and a second residual between the actual battery health status and the predicted battery health status are determined. Based on the first residual and the second residual, a preset anomaly detection algorithm is used to judge the first residual and the second residual, and the potential fault risk is assessed based on the judgment result. A battery status risk assessment result is generated based on the charging / discharging anomaly risk, the self-discharge anomaly risk, the capacity degradation risk, the start-up failure risk, and the potential fault risk.

[0011] In some embodiments, when assessing the charge / discharge abnormality risk of the battery based on the DC-DC output voltage, the DC-DC output current, and the battery voltage, the method includes: comparing the DC-DC output voltage and the DC-DC output current with preset output voltage safety thresholds and preset output current safety thresholds, respectively, to obtain a first comparison result; comparing the battery voltage with a preset battery voltage threshold to obtain a second comparison result; and assessing the charge / discharge abnormality risk of the battery based on the first comparison result and the second comparison result.

[0012] In some embodiments, when performing a risk assessment by calling the component evaluation model based on the vehicle component data to obtain the corresponding risk assessment result, the process includes: acquiring the control signal and status feedback signal of the vehicle component; assessing whether there is a control signal abnormality risk based on the amplitude, frequency, and waveform characteristics of the control signal; determining the engagement delay time based on the correspondence between the control signal and the status feedback signal, and assessing whether there is an engagement timing abnormality risk based on the engagement delay time; and generating a component status risk assessment result based on the control signal abnormality risk and the engagement timing abnormality risk.

[0013] In some embodiments, when performing a risk assessment based on the user behavior data and the ambient temperature data by calling the user behavior risk assessment model to obtain the corresponding risk assessment result, the process includes: acquiring historical user driving data; extracting vehicle usage intensity features, electricity usage intensity features, charging behavior features, and environmental correlation features from the historical user behavior data and the ambient temperature data to construct a user behavior feature vector; performing cluster analysis based on the user behavior feature vector using a preset learning algorithm to divide users into multiple user groups with similar behavior patterns; assigning a risk level and / or generating a risk score for each user group based on the historical vehicle power failure rate corresponding to each user group; and outputting the user behavior risk assessment result based on the risk level and / or risk score.

[0014] In some embodiments, predicting the risk level of a power-on failure of the vehicle based on the comprehensive risk assessment result includes: predicting the risk level of a power-on failure of the vehicle as a first risk level when the comprehensive risk assessment result is mapped to a first numerical range; predicting the risk level of a power-on failure of the vehicle as a second risk level when the comprehensive risk assessment result is mapped to a second numerical range; predicting the risk level of a power-on failure of the vehicle as a third risk level when the comprehensive risk assessment result is mapped to a third numerical range; and predicting the risk level of a power-on failure of the vehicle as a fourth risk level when the comprehensive risk assessment result is mapped to a fourth numerical range, wherein the first risk level is higher than the second risk level, the second risk level is higher than the third risk level, the third risk level is higher than the fourth risk level, and the first numerical range to the fourth numerical range do not overlap and cover the entire range of the comprehensive risk assessment result.

[0015] In some embodiments, when performing a risk assessment based on the vehicle operation data and the ambient temperature data by calling a preset assessment model corresponding to the vehicle component data, the battery data, and the user behavior data to obtain the corresponding risk assessment result, the method includes: grouping the vehicle operation data according to the ambient temperature data and the user behavior data to form multiple verification datasets; verifying the preset assessment model based on each verification dataset and continuously updating the vehicle operation data; and updating the preset assessment model according to the verification results and the updated vehicle operation data.

[0016] In some embodiments, after predicting the risk level of a power-on failure of the vehicle based on the comprehensive risk assessment results, the method includes: when the risk level is a first risk level, sending a first prompt message containing emergency alarm content to the user; when the risk level is a second risk level, sending a second prompt message containing risk reminders and usage optimization suggestions to the user; and when the risk level is a third risk level or a fourth risk level, generating an assessment report of normal vehicle status for the user to query.

[0017] To achieve the above objectives, a second aspect of the present invention discloses a prediction system for vehicle power-on failure, comprising: a data acquisition module for acquiring vehicle operating data and ambient temperature data, wherein the vehicle operating data includes vehicle component data, battery data, and user behavior data; an evaluation module for performing risk assessment based on the vehicle operating data and the ambient temperature data, by invoking a preset evaluation model corresponding to the vehicle component data, the battery data, and the user behavior data, and obtaining a corresponding risk assessment result, wherein the preset evaluation model includes a battery evaluation model, a component evaluation model, and a user behavior risk assessment model; a generation module for performing weighted fusion analysis on the risk assessment result to generate a comprehensive risk assessment result; and a prediction module for predicting the risk level of the vehicle power-on failure based on the comprehensive risk assessment result.

[0018] The vehicle power-on fault prediction system according to embodiments of the present invention, by collecting vehicle operation data and ambient temperature data, can comprehensively acquire key information reflecting the vehicle status and usage environment, effectively reducing the false alarm and false negative rates caused by traditional single-parameter judgment. Simultaneously, based on user behavior modeling, it performs refined analysis of different users' driving habits, achieving personalized risk assessment and providing accurate early warnings and customized suggestions, improving the targeting and effectiveness of management. Furthermore, based on vehicle operation data and ambient temperature data, it calls preset assessment models corresponding to vehicle component data, battery data, and user behavior data to conduct independent risk assessments from multiple dimensions, obtaining risk assessment results for each dimension. Subsequently, the risk assessment results for each dimension are normalized and weighted fusion analysis to generate a comprehensive risk assessment result. Finally, based on this comprehensive risk assessment result, the system predicts the risk level of a vehicle power-on fault. This allows for early identification of potential risks before a fault occurs, realizing a shift from passive maintenance to proactive early warning, providing users with sufficient response time, effectively avoiding vehicle interruption, and significantly improving user experience and overall vehicle reliability.

[0019] To achieve the above objectives, an embodiment of the third aspect of the present invention discloses a vehicle, comprising: a vehicle power-on fault prediction system as described in any embodiment of the second aspect of the present invention, or a processor, a memory, and a vehicle power-on fault prediction program stored in the memory and executable on the processor, wherein the vehicle power-on fault prediction program, when executed by the processor, implements the vehicle power-on fault prediction method as described in the embodiment of the second aspect of the present invention.

[0020] According to embodiments of the present invention, by collecting vehicle operation data and ambient temperature data, the vehicle can comprehensively acquire key information reflecting the vehicle's status and operating environment, effectively reducing the false alarm and false negative rates caused by traditional single-parameter judgment. Simultaneously, based on user behavior modeling, it performs refined analysis of different users' driving habits, achieving personalized risk assessment and providing accurate early warnings and customized suggestions, thus improving the targeting and effectiveness of management. Furthermore, based on vehicle operation data and ambient temperature data, it calls preset assessment models corresponding to vehicle component data, battery data, and user behavior data to conduct independent risk assessments from multiple dimensions, obtaining risk assessment results for each dimension. Subsequently, the risk assessment results for each dimension are normalized and weighted fusion analysis to generate a comprehensive risk assessment result. Finally, based on this comprehensive risk assessment result, the risk level of a vehicle power-on failure is predicted. This allows for the early identification of potential risks before a failure occurs, realizing a shift from passive maintenance to proactive early warning, providing users with sufficient response time, effectively avoiding vehicle interruptions, and significantly improving user experience and overall vehicle reliability.

[0021] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0022] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which: Figure 1 This is a flowchart of a method for predicting vehicle power-on faults according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the relationship between the K value and time according to an embodiment of the present invention; Figure 3 This is a schematic diagram of a reliability fitting curve according to an embodiment of the present invention; Figure 4 This is a structural block diagram of a vehicle power-on fault prediction system according to an embodiment of the present invention; Figure 5 This is a structural block diagram of a vehicle according to an embodiment of the present invention; Figure 6 This is a structural block diagram of a vehicle according to another embodiment of the present invention. Detailed Implementation

[0023] The embodiments of the present invention are described in detail below. The embodiments described with reference to the accompanying drawings are exemplary. The embodiments of the present invention are described in detail below.

[0024] The following is for reference. Figures 1-3 A method for predicting vehicle power-on faults according to embodiments of the present invention is described.

[0025] like Figure 1 The diagram shows a flowchart of a method for predicting vehicle power-on faults according to an embodiment of the present invention. The method includes at least steps S1-S4.

[0026] Step S1: Collect vehicle operation data and ambient temperature data. Vehicle operation data includes vehicle component data, battery data, and user behavior data.

[0027] In this embodiment, vehicle-mounted sensors (including but not limited to battery voltage sensors, current sensors, and environmental and battery temperature sensors), electronic control units (such as motor control units and battery management systems), and remote information processors work together to collect vehicle operation data and environmental temperature data in real time. The data is then aggregated through the vehicle communication network and uploaded to the cloud platform for subsequent power-on fault risk prediction.

[0028] The vehicle operation data encompasses three main categories: battery data, vehicle component data, and user behavior data. Ambient temperature data includes real-time outside temperature and battery compartment temperature obtained through onboard ambient temperature sensors, as well as historical, current, and future predicted ambient temperatures obtained by integrating vehicle location information with external weather services. This data is used to analyze the impact of temperature on battery performance and user behavior. Component data primarily refers to the operating status of core relays closely related to power supply, including relay engagement status feedback signals reported by the body control module and relay control command signals issued by the central control unit. This data is used to diagnose relay malfunctions such as response delays, contact sticking, or control failures. User behavior data includes daily mileage, cumulative driving time, average speed, frequency of rapid acceleration and deceleration, number of daily starts, and the activation time, duration, and power consumption level of in-vehicle electrical appliances after engine shutdown. It also records charging-related behaviors, such as the start and end times, charging duration, charging power, battery state of charge at the start and end of each charge, and charging location information, comprehensively depicting the user's vehicle usage, power consumption, and charging habits. These data form the basic inputs for the pre-set evaluation model, enabling accurate, forward-looking, and personalized predictions of vehicle power-on faults.

[0029] Step S2: Based on vehicle operation data and ambient temperature data, call the preset assessment models corresponding to vehicle component data, battery data and user behavior data to conduct risk assessment and obtain the corresponding risk assessment results. The preset assessment models include battery assessment model, component assessment model and user behavior risk assessment model.

[0030] In this embodiment, based on the collected vehicle operation data and ambient temperature data, a preset assessment model corresponding to the vehicle component data, battery data, and user behavior data is invoked to conduct a multi-dimensional risk assessment, and the risk assessment results for each dimension are obtained.

[0031] For example, the battery assessment model comprehensively evaluates the battery's health status and risk of depletion based on the battery's voltage decay characteristics during rest, DC-DC output status, intelligent charging behavior, and voltage trend prediction. The component assessment model analyzes the potential for faults such as coil aging, contact adhesion, and control circuit abnormalities based on the time response relationship, waveform characteristics, and electrical parameters of the control command signals and actual engagement signals of key relays. The user behavior risk assessment model extracts behavioral characteristics and constructs user profiles based on factors such as user driving frequency, parking duration, post-engine shutdown charging habits, charging patterns, and the influence of ambient temperature, identifying high-risk vehicle usage patterns. Thus, based on the synergistic analysis of these three assessment models, risk assessment results are output for the battery system, key components, and user behavior, providing a quantitative basis for subsequent comprehensive risk fusion and power-on fault prediction.

[0032] Step S3: Perform a weighted fusion analysis on the risk assessment results to generate a comprehensive risk assessment result.

[0033] In this embodiment, the risk assessment results of each dimension are weighted and fused for analysis, that is, a comprehensive risk assessment result is generated based on the risk assessment results of battery status, component status, and user behavior.

[0034] Specifically, the results of the battery status risk assessment, the component status risk assessment, and the user behavior risk assessment are integrated to obtain a precise comprehensive risk assessment result. Each assessment result is normalized to a score range of 0-100, with lower scores indicating higher risks in the corresponding dimension.

[0035] For example, a weighted linear fusion algorithm can be used to calculate the comprehensive risk assessment result. The comprehensive risk assessment result can be denoted as S_total, the battery status risk assessment result as S_batt, the component status risk assessment result as S_relay, and the user behavior risk assessment result as S_user.

[0036] The comprehensive risk assessment result S_total is calculated as follows: S_total = W_batt × S_batt + W_user × S_user + W_relay × S_relay, where W_batt, W_user, and W_relay are the weight coefficients of the corresponding risk factors, and the sum of the weights is 1 (i.e., W_batt + W_user + W_relay = 1).

[0037] Furthermore, the weight allocation is determined based on the actual impact of each factor on the vehicle's inability to power on. For example, the battery status, as the most direct determining factor, is given the highest weight (e.g., 0.5), user behavior, as a long-term cumulative risk factor, is given a relatively large weight (e.g., 0.4), and the status of components (e.g., relay status), as a sudden but serious fault factor, is given a moderate weight (e.g., 0.1). Each weight parameter can be trained and optimized based on large-scale historical fault data through statistical modeling or machine learning methods. The final comprehensive risk assessment result S_total is used as the system-level decision output, reflecting the overall power-on risk level of the vehicle, and realizing the transformation from multi-source heterogeneous data to a unified risk criterion.

[0038] Step S4: Based on the comprehensive risk assessment results, predict the risk level of the vehicle's power-on failure.

[0039] In this embodiment, the comprehensive risk assessment result S_total obtained by weighted fusion calculation is mapped to a preset discrete risk level range to achieve a graded judgment of the vehicle power-on failure risk. The comprehensive risk assessment result S_total comprehensively reflects the coupled influence of battery health status, user behavior risk, and key component (such as relay) status. The lower the score, the higher the overall risk. By converting continuous comprehensive scores into intuitive levels such as high risk, medium risk, and low risk, this mechanism transforms complex assessment results into interpretable decision outputs, which facilitates triggering differentiated early warning notifications, service interventions, or remote diagnostic strategies, and realizes closed-loop management from risk identification to proactive response.

[0040] Therefore, the aforementioned method for predicting vehicle power-on failures, by collecting vehicle operation data and ambient temperature data, can comprehensively acquire key information reflecting the vehicle's status and operating environment. This effectively reduces the false alarm and false negative rates caused by traditional single-parameter judgments. Simultaneously, based on user behavior modeling, it performs refined analysis of different users' driving habits, achieving personalized risk assessment and providing accurate early warnings and customized suggestions, thus improving the targeting and effectiveness of management. Furthermore, based on vehicle operation data and ambient temperature data, it calls preset assessment models corresponding to vehicle component data, battery data, and user behavior data to conduct independent risk assessments from multiple dimensions, obtaining risk assessment results for each dimension. Subsequently, the risk assessment results for each dimension are normalized and weighted fusion analysis to generate a comprehensive risk assessment result. Finally, based on this comprehensive risk assessment result, the risk level of a vehicle power-on failure is predicted. This allows for the early identification of potential risks before a failure occurs, realizing a shift from passive maintenance to proactive early warning, providing users with sufficient response time, effectively avoiding vehicle interruptions, and significantly improving user experience and overall vehicle reliability.

[0041] In one embodiment of the present invention, the battery data includes the vehicle's stationary voltage data, DC-DC output voltage, DC-DC output current, and the vehicle's battery voltage during the vehicle's resting period. When performing a risk assessment based on the battery data and ambient temperature data, and obtaining the corresponding risk assessment result, the process includes: assessing the battery's charge / discharge abnormality risk based on the DC-DC output voltage, DC-DC output current, and battery voltage; predicting the natural charge loss rate per unit time using a preset regression algorithm based on the stationary voltage data and ambient temperature data, and assessing the battery's self-discharge abnormality risk based on the natural charge loss rate; obtaining the battery's power-off time and the number of times the battery was recharged during the power-off period; establishing a mapping relationship between the recharge frequency and the battery's health status using a preset reliability distribution model based on the power-off time, the number of recharges, and the ambient temperature data; and predicting the current battery health status value based on the mapping relationship to assess the battery's health status. The risk of battery capacity degradation is assessed through the following steps: Time-series data is acquired, and a preset time-series analysis algorithm is used to predict the battery voltage change trend. When the predicted battery voltage is lower than a preset power-on failure threshold, a start-up failure risk is assessed. The time-series data includes historical battery voltage data, historical ambient temperature data, and historical user behavior data. The actual battery self-discharge rate and actual battery health status value are obtained. A first residual is determined between the actual and predicted self-discharge rates, and a second residual is determined between the actual and predicted battery health status. Based on the first and second residuals, a preset anomaly detection algorithm is used to judge the first and second residuals, and the potential fault risk is assessed based on the judgment results. A battery status risk assessment result is generated based on the risks of abnormal charging and discharging, abnormal self-discharge, capacity degradation, start-up failure, and potential faults.

[0042] In this embodiment, the battery data includes, but is not limited to, the stationary voltage data of the vehicle during the resting period, the DC-DC output voltage, the DC-DC output current, and the battery voltage. The battery voltage includes the battery voltage at the time of intelligent charging triggering and termination, as well as the battery voltage collected by the control unit when the vehicle is powered on.

[0043] Based on the aforementioned battery and ambient temperature data, a battery assessment model is invoked to conduct a risk assessment, comprehensively analyzing the battery's health status and operational risks from multiple dimensions. First, the output voltage and current of the DC-DC converter are monitored to determine if they exceed preset safety thresholds. This is combined with monitoring whether the battery voltage under different operating conditions falls below a discharge threshold. For example, a battery voltage below 12.1V indicates a risk of discharge, below 11V indicates severe discharge, and above the overcharge threshold of 14V indicates an overcharge risk. Based on these assessments, it is determined whether the battery exhibits abnormal charging or over-discharging behavior, thus evaluating the risk of abnormal charging and discharging.

[0044] Secondly, using battery voltage decay data during the vehicle's idle period after being turned off, combined with ambient temperature information, a self-discharge model is established using a preset regression algorithm (such as linear regression or random forest regression) to calculate the natural energy loss rate (i.e., K value) per unit time. Figure 2 The diagram showing the relationship between K value and time indicates that when the loss rate is significantly higher than the normal level, it suggests that the battery has an internal micro-short circuit or increased self-discharge. Based on this, the risk of abnormal self-discharge can be assessed.

[0045] The duration of each vehicle power outage and the number of times the intelligent charging function was triggered during this period were statistically analyzed. Combined with ambient temperature data, a pre-defined reliability distribution model (such as the Weibull distribution) was used to establish a mapping relationship between the charging frequency and the battery health status. Figure 3 The reliability fitting curve diagram shown estimates the current battery health status. An abnormally increased frequency of recharging indicates battery capacity degradation, used to assess the risk of capacity degradation. Simultaneously, based on a time series of historical battery voltage, historical ambient temperature, and historical user behavior data, time series analysis algorithms, such as ARIMA (Autoregressive Integrated Moving Average Model) and exponential smoothing, are used to predict the battery voltage trend over a future period. When the predicted voltage falls below the threshold preventing the vehicle from powering on, a risk of start-up failure is identified.

[0046] In addition, the actual natural power loss rate and actual battery health status value are obtained, and the first residual and the second residual between them and the model prediction value are calculated respectively. These are then input into a preset anomaly detection algorithm (such as isolated forest or local outlier factor) for judgment. If the residual deviates significantly from the normal range, it is identified as abnormal behavior and used to assess the potential failure risks of the battery body, battery management system or other related components.

[0047] Finally, the risks of abnormal charging and discharging, abnormal self-discharge, capacity degradation, start-up failure, and potential faults output by the five sub-models are normalized and converted into intermediate scores with uniform dimensions (e.g., 0-100 points). Weighting coefficients are then assigned based on the contribution of each risk dimension to the overall fault, and a weighted fusion algorithm is used to generate the final battery status score. This score comprehensively reflects the current overall health level of the battery; a lower score indicates poorer battery performance and a higher risk of power-on failure. This battery-level risk assessment result is then used for further integration with other risk dimensions such as user behavior and key components, achieving accurate, multi-dimensional, and quantifiable prediction and assessment of vehicle power-on risks.

[0048] In one embodiment of the present invention, when assessing the risk of abnormal charging and discharging of a battery based on the DC-DC output voltage, DC-DC output current, and battery voltage, the method includes: comparing the DC-DC output voltage and DC-DC output current with preset output voltage safety thresholds and preset output current safety thresholds, respectively, to obtain a first comparison result; comparing the battery voltage with a preset battery voltage threshold to obtain a second comparison result; and assessing the risk of abnormal charging and discharging of the battery based on the first comparison result and the second comparison result.

[0049] In this embodiment, the real-time acquired DC-DC converter output voltage is compared with a preset output voltage safety threshold. If the value is lower than the lower threshold (e.g., 11.5V) or higher than the upper threshold (e.g., 14.5V), an abnormal output voltage is determined, which may manifest as insufficient charging or overcharging risk. Simultaneously, the DC-DC output current is compared with a preset output current safety threshold. If the current continuously exceeds the rated range (e.g., exceeding 10A), it indicates that the low-voltage system is overloaded or there is abnormal power consumption, which may indicate abnormal discharge or component short circuit risk. This yields the first comparison result. Secondly, the vehicle's battery voltage under different operating conditions is compared with a preset battery voltage threshold. For example, when the voltage is below 12.1V, a slight low battery level is indicated. The system assesses the risks of battery depletion. A voltage below 11V indicates severe undercharging, posing a risk of failure to start the battery. A voltage above 14.0V may indicate overcharging, potentially leading to water loss or accelerated aging. This results in a second comparison. Finally, a comprehensive judgment is made by combining the first and second comparison results. If the DC-DC output is abnormal and the battery voltage remains low, it may indicate a charging system malfunction causing insufficient charging. If the DC-DC output is normal but the battery voltage drops rapidly, it may indicate battery capacity degradation or a leakage path. If both are abnormal, it may indicate a control logic error or a battery management system malfunction. This multi-dimensional voltage and current comparison logic accurately identifies abnormal operating conditions during charging and discharging, thereby assessing the risk of abnormal charging and discharging of the battery.

[0050] In one embodiment of the present invention, when performing risk assessment by calling a component evaluation model based on vehicle component data to obtain the corresponding risk assessment result, the process includes: acquiring control signals and status feedback signals of the vehicle components; assessing whether there is a risk of control signal abnormality based on the amplitude, frequency, and waveform characteristics of the control signals; determining the engagement delay time based on the correspondence between the control signals and the status feedback signals, and assessing whether there is a risk of engagement timing abnormality based on the engagement delay time; and generating a component status risk assessment result based on the control signal abnormality risk and the engagement timing abnormality risk.

[0051] In this embodiment, when calling the component evaluation model to perform risk assessment based on vehicle component data, for example, taking a critical power relay as the evaluation object, the electrical and mechanical performance of the relay is diagnosed in multiple dimensions by collecting its control signals and status feedback signals, thereby generating a component status risk assessment result.

[0052] Specifically, the process begins by acquiring the relay's control signal and status feedback signal. Analysis of the control signal's amplitude, frequency, and waveform characteristics assesses the potential for abnormal control signal performance. During the high-level period (i.e., when the activation command is issued), the amplitude and frequency characteristics are extracted. If the amplitude is significantly lower than the standard value (e.g., below 10V), it may indicate excessive contact resistance or insufficient power supply in the control circuit. An abnormally high amplitude suggests a control module malfunction. If the frequency deviates from the rated range, it may affect the relay's stable activation, causing contact chatter. Simultaneously, the waveform of the control signal is analyzed. Distortions such as tilting or steps on the rising or falling edges may reflect decreased driving capability or circuit aging. High-frequency glitches or noise detected during the high-level period may be caused by electromagnetic interference or an internal control unit malfunction, thus determining whether the control signal is abnormal.

[0053] Secondly, based on the timing correspondence between the control signal and the status feedback signal, the pull-in delay time T_delay is calculated, which is the time interval between the rising edge of the control signal and the rising edge of the status feedback signal, and the risk of abnormal pull-in timing is assessed. If the measured pull-in delay time T_delay is significantly longer than the nominal value or the historical normal range, it may indicate that the relay coil is aging, mechanically stuck, or the power supply capacity is reduced. If the pull-in delay time T_delay is 0 or extremely short, it may be due to contact sticking leading to normal closure. If there is no response to the status signal, it may be due to coil open circuit, mechanical failure, or feedback circuit failure.

[0054] Finally, control signal over-limits, abnormal engagement time, and waveform distortion are converted into quantitative indicators, such as Boolean anomaly flags or continuous anomaly scores based on the degree of deviation, and then normalized to a unified intermediate score (0-100 points). Subsequently, weights are assigned based on the contribution of each dimension to relay failure, and a weighted fusion algorithm generates the final relay status score. A lower score indicates a higher probability of physical aging, contact adhesion, coil short circuit, or control circuit failure in the relay. This component status risk assessment comprehensively reflects the health level of key relays and serves as a risk output at the component level, used for subsequent integrated analysis with dimensions such as battery and user behavior to achieve accurate prediction of vehicle power-on failures.

[0055] In one embodiment of the present invention, when performing a risk assessment based on user behavior data and ambient temperature data by calling a user behavior risk assessment model to obtain the corresponding risk assessment result, the process includes: acquiring historical user driving data; extracting vehicle usage intensity features, electricity usage intensity features, charging behavior features, and environmental correlation features from the historical user behavior data and ambient temperature data to construct a user behavior feature vector; performing cluster analysis based on the user behavior feature vector using a preset learning algorithm to divide users into multiple user groups with similar behavior patterns; assigning a risk level and / or generating a risk score for each user group based on the historical vehicle power failure rate corresponding to each user group; and outputting the user behavior risk assessment result based on the risk level and / or risk score.

[0056] In this embodiment, when using a user behavior risk assessment model based on user behavior data and ambient temperature data to perform risk assessment, historical user driving data is first acquired, and multi-dimensional behavioral features are extracted from it to construct a comprehensive user behavior feature vector. Specifically, this includes: vehicle usage intensity features, such as average daily mileage, average daily driving time, average daily start-up frequency, and frequency of rapid acceleration / deceleration, used to distinguish between high-frequency long-distance driving, mid-frequency regional driving, and low-frequency sporadic driving modes; electricity consumption intensity features, such as average daily usage time of electrical appliances after parking and average power consumption level, used to identify users who engage in high-power-consuming behaviors such as prolonged use of air conditioning, central control screen, lights, or audio after the engine is turned off; charging behavior features, such as average charging interval days, average initial state of charge, and DC fast charging usage ratio, used to classify charging habits such as timely charging, on-demand charging, and passive charging; and environmental correlation features, such as the proportion of average daily mileage in low-temperature environments, used to capture the amplified risk effect of users' short-distance commutes in cold climates leading to insufficient battery charging.

[0057] Based on the aforementioned feature vectors, a pre-defined learning algorithm is used to perform cluster analysis on users, dividing them into several groups with similar behavioral patterns (e.g., 3×3×3=27 groups) according to multiple dimensions such as vehicle usage time, charging habits, and ambient temperature. Subsequently, combined with the historical vehicle power failure rate corresponding to each user group, the risk exposure level of different groups is calculated, and each group is assigned a corresponding risk level (e.g., high risk, medium risk, low risk) and / or a quantitative risk score (e.g., high risk 30 points, medium risk 60 points, low risk 85 points, with lower scores indicating higher risk).

[0058] Ultimately, user behavior risk assessment results are generated based on the risk level and / or risk score, enabling differentiated risk identification for users with different driving habits, providing personalized input for subsequent comprehensive risk integration, and improving the accuracy and interpretability of power-on fault prediction.

[0059] In one embodiment of the present invention, when predicting the risk level of a vehicle power-on failure based on a comprehensive risk assessment result, the method includes: when the comprehensive risk assessment result is mapped to a first numerical range, the predicted risk level of the vehicle power-on failure is a first risk level; when the comprehensive risk assessment result is mapped to a second numerical range, the predicted risk level of the vehicle power-on failure is a second risk level; when the comprehensive risk assessment result is mapped to a third numerical range, the predicted risk level of the vehicle power-on failure is a third risk level; and when the comprehensive risk assessment result is mapped to a fourth numerical range, the predicted risk level of the vehicle power-on failure is a fourth risk level. The first risk level is higher than the second risk level, the second risk level is higher than the third risk level, the third risk level is higher than the fourth risk level, and the first to fourth numerical ranges do not overlap and cover the entire range of the comprehensive risk assessment result.

[0060] In this embodiment, when predicting the risk level of a vehicle power-on failure based on the comprehensive risk assessment results, the risk level is classified and determined by mapping the comprehensive risk assessment results to a preset discrete level range.

[0061] Specifically, when the comprehensive risk assessment result is within the first numerical range (0 ≤ S_total < 40), the predicted risk level for a vehicle power-on failure is the first risk level, i.e., high risk, indicating that the vehicle has an extremely high probability of being unable to power on in the short term (e.g., within 24 hours), requiring immediate intervention. When the comprehensive risk assessment result is within the second numerical range (40 ≤ S_total < 60), the predicted risk level is the second risk level, i.e., medium risk, indicating significant risk factors exist, and users should pay attention and are advised to optimize their vehicle usage behavior. When the comprehensive risk assessment result is within the third numerical range (60 ≤ S_total < 80), the predicted risk level is the third risk level, i.e., low risk, indicating that the overall condition of the vehicle is controllable, the risk is low, and continued monitoring and operation are possible. When the comprehensive risk assessment result is within the fourth numerical range (80 ≤ S_total ≤ 100), the predicted risk level is the fourth risk level, i.e., excellent, indicating that the battery, key components, and user behavior are all in good condition, with no significant risk of power-on failure.

[0062] In this system, the first risk level is higher than the second risk level, the second risk level is higher than the third risk level, and the third risk level is higher than the fourth risk level. Furthermore, the numerical ranges of the first to fourth risk levels do not overlap, collectively covering the full range of the comprehensive risk assessment result (i.e., 0-100 points). This grading mechanism transforms continuous comprehensive risk assessment results into intuitive and operable discrete risk levels, facilitating the system's subsequent triggering of differentiated early warning strategies and service responses, thus achieving closed-loop management from quantitative assessment to practical application.

[0063] In one embodiment of the present invention, when performing a risk assessment based on vehicle operation data and ambient temperature data by calling a preset assessment model corresponding to vehicle component data, battery data, and user behavior data, and obtaining the corresponding risk assessment result, the method includes: grouping the vehicle operation data according to ambient temperature data and user behavior data to form multiple verification datasets; verifying the preset assessment model based on each verification dataset and continuously updating the vehicle operation data; and updating the preset assessment model according to the verification results and the updated vehicle operation data.

[0064] In this embodiment, vehicle operation data is grouped in multiple dimensions based on ambient temperature data and user behavior data. For example, it is divided into multiple verification datasets based on temperature ranges such as low temperature, normal temperature, and high temperature, as well as user behavior patterns such as high-frequency long-distance travel, low-frequency short-distance travel, and high power consumption, to cover typical operating conditions under different climate conditions and vehicle usage scenarios. Subsequently, the corresponding preset evaluation model is validated based on each grouped dataset.

[0065] Specifically, for battery evaluation models (such as self-discharge rate prediction and voltage trend prediction), the prediction accuracy is evaluated by calculating the root mean square error between the predicted and measured values. For user behavior risk assessment models, the rationality and stability of clustering results are evaluated by using indicators such as classification accuracy and F1-score. At the same time, using historical data of vehicles that have experienced power failures, the outlier detection models in the battery evaluation models, such as residual-based isolated forests or LOF (Local Outlier Factor) algorithms, are backtested to verify whether they can effectively identify abnormal patterns before the failure occurs, thereby confirming the actual discrimination ability of the model.

[0066] Based on this, vehicle operation data is continuously collected and updated to form a dynamically growing data pool. According to the verification feedback results of each model and the continuously accumulated new data, the model is iteratively optimized to achieve continuous improvement in the accuracy, robustness and adaptability of the evaluation model. Ultimately, the model achieves self-evolution and adaptive performance updates, ensuring that the risk assessment results always maintain high reliability and predictive ability in diverse operating conditions and user groups.

[0067] In one embodiment of the present invention, after predicting the risk level of a vehicle power-on failure based on a comprehensive risk assessment, the method includes: when the risk level is a first risk level, sending a first prompt message containing emergency alarm content to the user; when the risk level is a second risk level, sending a second prompt message containing risk reminders and usage optimization suggestions to the user; and when the risk level is a third or fourth risk level, generating an assessment report of normal vehicle status for the user to query.

[0068] In this embodiment, after predicting the risk level of a vehicle power-on failure based on a comprehensive risk assessment, the system executes differentiated warning and service strategies according to different risk levels.

[0069] Specifically, when the risk level is Level 1 (high risk), the system determines that the vehicle has an extremely high risk of short-term power-on failure. The system immediately triggers the highest priority alarm and sends a first prompt message containing emergency alarm content to the user terminal. This can be proactively reached through various channels such as user terminal push notifications, SMS notifications, and in-vehicle voice broadcasts. The prompt message includes a clear explanation of the fault risk and emergency operation guidelines, such as: if the battery is severely depleted and there is a risk of not being able to start, please avoid using electricity after turning off the engine and immediately drive to charge for more than 30 minutes. When the risk level is Level 2 (medium risk), the system generates a second prompt message containing risk reminders and usage optimization suggestions. This message is pushed through non-emergency channels such as the user terminal message center and in-vehicle information screens to suggest improvements to the user's driving habits. For example, if your vehicle has been parked for a long time recently, it is recommended to drive for a longer period of time to fully recharge the battery. When the risk level is Level 3 (low risk) or Level 4 (excellent), the system does not send a proactive warning push, but generates a comprehensive assessment report on the normal status of the vehicle's low-voltage system. Users can check the report themselves through their user terminals to understand the vehicle's health status.

[0070] The aforementioned risk levels are pushed to different service applications through a standardized API (Application Programming Interface). For example, for individual users, warning information and service suggestions are delivered precisely through mobile apps, SMS, and in-vehicle terminals. For enterprise users (such as fleet operators), the system aggregates and pushes overall risk distribution, high-risk vehicle lists, driver's poor electrical behavior (such as frequent deep discharge) reminders, and management decision support information such as backup power configuration suggestions in low-temperature weather through the fleet management platform. This achieves comprehensive coverage from individuals to groups and from warnings to management, improving user service experience and vehicle operation reliability.

[0071] The vehicle power-on fault prediction method according to embodiments of the present invention, by collecting vehicle operation data and ambient temperature data, can comprehensively acquire key information reflecting the vehicle status and usage environment, effectively reducing the false alarm and false negative rates caused by traditional single-parameter judgment. Simultaneously, based on user behavior modeling, it performs refined analysis of different users' driving habits, achieving personalized risk assessment and providing accurate early warnings and customized suggestions, improving the targeting and effectiveness of management. Based on this, based on vehicle operation data and ambient temperature data, it calls preset assessment models corresponding to vehicle component data, battery data, and user behavior data respectively, conducting independent risk assessments from multiple dimensions to obtain risk assessment results for each dimension. Subsequently, the risk assessment results for each dimension are normalized and weighted fusion analysis to generate a comprehensive risk assessment result. Finally, based on this comprehensive risk assessment result, the risk level of a vehicle power-on fault is predicted. This allows for early identification of potential risks before a fault occurs, realizing a shift from passive maintenance to proactive early warning, providing users with sufficient response time, effectively avoiding vehicle interruption, and significantly improving user experience and overall vehicle reliability.

[0072] A further embodiment of the present invention discloses a predictive system for vehicle power-on failures.

[0073] like Figure 4 The diagram shown is a structural block diagram of a vehicle power-on fault prediction system according to an embodiment of the present invention. The vehicle power-on fault prediction system 100 includes: a data acquisition module 110, an evaluation module 120, a generation module 130, and a prediction module 140.

[0074] The system includes: a data acquisition module 110 for collecting vehicle operation data and ambient temperature data, including vehicle component data, battery data, and user behavior data; an evaluation module 120 for using the vehicle operation data and ambient temperature data to perform risk assessments by calling preset evaluation models corresponding to the vehicle component data, battery data, and user behavior data, and obtaining corresponding risk assessment results, including battery evaluation models, component evaluation models, and user behavior risk assessment models; a generation module 130 for performing weighted fusion analysis on the risk assessment results to generate a comprehensive risk assessment result; and a prediction module 140 for predicting the risk level of a vehicle power-on failure based on the comprehensive risk assessment results.

[0075] In one embodiment of the present invention, the battery data includes the vehicle's resting voltage data during the resting period, the DC-DC output voltage, the DC-DC output current, and the vehicle's battery voltage. When performing a risk assessment based on the battery data and ambient temperature data, and obtaining the corresponding risk assessment result, the following steps are included: Assess the risk of abnormal charging and discharging of the battery based on the DC-DC output voltage, DC-DC output current, and battery voltage. Based on static voltage and ambient temperature data, a preset regression algorithm is used to predict the natural power loss rate per unit time, and the risk of battery self-discharge is assessed based on the natural power loss rate. The battery's power-off time and the number of times it was recharged during the power-off period are obtained. Based on the power-off time, recharge frequency, and ambient temperature data, a preset reliability distribution model is used to establish a mapping relationship between recharge frequency and battery health status. The current battery health status value is predicted based on this mapping relationship to assess the risk of battery capacity degradation. Time series data is obtained, and a preset time series analysis algorithm is used to predict the battery voltage change trend. When the predicted battery voltage is lower than a preset power-off threshold, the possibility of starting is assessed. Failure risk assessment includes time-series data such as historical battery voltage data, historical ambient temperature data, and historical user behavior data; obtaining the actual natural battery depletion rate and the actual battery health status value; determining the first residual between the actual and predicted natural battery depletion rates; determining the second residual between the actual and predicted battery health status; using a preset anomaly detection algorithm to judge the first and second residuals based on the judgment results; and generating a battery status risk assessment result based on the risks of abnormal charging and discharging, abnormal self-discharge, capacity degradation, start-up failure, and potential failures.

[0076] In one embodiment of the present invention, when assessing the risk of abnormal charging and discharging of a battery based on the DC-DC output voltage, DC-DC output current, and battery voltage, the method includes: comparing the DC-DC output voltage and DC-DC output current with preset output voltage safety thresholds and preset output current safety thresholds, respectively, to obtain a first comparison result; comparing the battery voltage with a preset battery voltage threshold to obtain a second comparison result; and assessing the risk of abnormal charging and discharging of the battery based on the first comparison result and the second comparison result.

[0077] In one embodiment of the present invention, when performing risk assessment by calling a component evaluation model based on vehicle component data to obtain the corresponding risk assessment result, the process includes: acquiring control signals and status feedback signals of the vehicle components; assessing whether there is a risk of control signal abnormality based on the amplitude, frequency, and waveform characteristics of the control signals; determining the engagement delay time based on the correspondence between the control signals and the status feedback signals, and assessing whether there is a risk of engagement timing abnormality based on the engagement delay time; and generating a component status risk assessment result based on the control signal abnormality risk and the engagement timing abnormality risk.

[0078] In one embodiment of the present invention, when performing a risk assessment based on user behavior data and ambient temperature data by calling a user behavior risk assessment model to obtain the corresponding risk assessment result, the process includes: acquiring historical user driving data; extracting vehicle usage intensity features, electricity usage intensity features, charging behavior features, and environmental correlation features from the historical user behavior data and ambient temperature data to construct a user behavior feature vector; performing cluster analysis based on the user behavior feature vector using a preset learning algorithm to divide users into multiple user groups with similar behavior patterns; assigning a risk level and / or generating a risk score for each user group based on the historical vehicle power failure rate corresponding to each user group; and outputting the user behavior risk assessment result based on the risk level and / or risk score.

[0079] In one embodiment of the present invention, when predicting the risk level of a vehicle power-on failure based on a comprehensive risk assessment result, the method includes: when the comprehensive risk assessment result is mapped to a first numerical range, the predicted risk level of the vehicle power-on failure is a first risk level; when the comprehensive risk assessment result is mapped to a second numerical range, the predicted risk level of the vehicle power-on failure is a second risk level; when the comprehensive risk assessment result is mapped to a third numerical range, the predicted risk level of the vehicle power-on failure is a third risk level; and when the comprehensive risk assessment result is mapped to a fourth numerical range, the predicted risk level of the vehicle power-on failure is a fourth risk level. The first risk level is higher than the second risk level, the second risk level is higher than the third risk level, the third risk level is higher than the fourth risk level, and the first to fourth numerical ranges do not overlap and cover the entire range of the comprehensive risk assessment result.

[0080] In one embodiment of the present invention, when performing a risk assessment based on vehicle operation data and ambient temperature data by calling a preset assessment model corresponding to vehicle component data, battery data, and user behavior data, and obtaining the corresponding risk assessment result, the method includes: grouping the vehicle operation data according to ambient temperature data and user behavior data to form multiple verification datasets; verifying the preset assessment model based on each verification dataset and continuously updating the vehicle operation data; and updating the preset assessment model according to the verification results and the updated vehicle operation data.

[0081] In one embodiment of the present invention, after predicting the risk level of a vehicle power-on failure based on a comprehensive risk assessment, the method includes: when the risk level is a first risk level, sending a first prompt message containing emergency alarm content to the user; when the risk level is a second risk level, sending a second prompt message containing risk reminders and usage optimization suggestions to the user; and when the risk level is a third or fourth risk level, generating an assessment report of normal vehicle status for the user to query.

[0082] The vehicle power-on fault prediction system 100 according to an embodiment of the present invention can comprehensively acquire key information reflecting the vehicle status and usage environment by collecting vehicle operation data and ambient temperature data, effectively reducing the false alarm and false negative rates caused by traditional single-parameter judgment. Simultaneously, based on user behavior modeling, it performs refined analysis of different users' driving habits, achieving personalized risk assessment and providing accurate early warnings and customized suggestions, improving the targeting and effectiveness of management. Based on this, based on vehicle operation data and ambient temperature data, it calls preset assessment models corresponding to vehicle component data, battery data, and user behavior data to conduct independent risk assessments from multiple dimensions, obtaining risk assessment results for each dimension. Subsequently, the risk assessment results for each dimension are normalized and weighted fusion analysis to generate a comprehensive risk assessment result. Finally, based on this comprehensive risk assessment result, the risk level of a vehicle power-on fault is predicted. This allows for early identification of potential risks before a fault occurs, realizing a shift from passive maintenance to proactive early warning, providing users with sufficient response time, effectively avoiding vehicle interruption, and significantly improving user experience and overall vehicle reliability.

[0083] A further embodiment of the present invention discloses a vehicle.

[0084] In some embodiments, such as Figure 5 As shown, vehicle 200 includes the vehicle power-on fault prediction system 100 described in the above embodiments of the present invention.

[0085] In other embodiments, such as Figure 6 As shown, the vehicle 200 includes a processor 201, a memory 202, and a vehicle power-on fault prediction program stored in the memory 202 and executable on the processor 201. When the vehicle power-on fault prediction program is executed by the processor 201, it implements the vehicle violation information prompting and vehicle power-on fault prediction method as described in the above embodiments of the present invention.

[0086] According to an embodiment of the present invention, the vehicle 200, by collecting vehicle operation data and ambient temperature data, can comprehensively acquire key information reflecting the vehicle's status and usage environment, effectively reducing the false alarm and false negative rates caused by traditional single-parameter judgment. Simultaneously, based on user behavior modeling, it performs refined analysis of different users' driving habits, achieving personalized risk assessment and providing accurate early warnings and customized suggestions, thus improving the targeting and effectiveness of management. Furthermore, based on vehicle operation data and ambient temperature data, it calls preset assessment models corresponding to vehicle component data, battery data, and user behavior data to conduct independent risk assessments from multiple dimensions, obtaining risk assessment results for each dimension. Subsequently, the risk assessment results for each dimension are normalized and weighted fusion analysis to generate a comprehensive risk assessment result. Finally, based on this comprehensive risk assessment result, the risk level of a vehicle power-on failure is predicted. This allows for the early identification of potential risks before a failure occurs, realizing a shift from passive maintenance to proactive early warning, providing users with sufficient response time, effectively avoiding vehicle interruption, and significantly improving user experience and overall vehicle reliability.

[0087] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example.

[0088] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for predicting vehicle power-on faults, characterized in that, include: Collect vehicle operation data and ambient temperature data. The vehicle operation data includes vehicle component data, battery data, and user behavior data. Based on the vehicle operation data and the ambient temperature data, a preset assessment model corresponding to the vehicle component data, the battery data and the user behavior data is invoked to perform a risk assessment and obtain the corresponding risk assessment result. The preset assessment model includes a battery assessment model, a component assessment model and a user behavior risk assessment model. The risk assessment results are weighted and fused to generate a comprehensive risk assessment result; Based on the comprehensive risk assessment results, the risk level of the vehicle experiencing a power-on failure is predicted.

2. The method for predicting vehicle power-on faults according to claim 1, characterized in that, The battery data includes the vehicle's static voltage data, DC-DC output voltage, DC-DC output current, and the vehicle's battery voltage during the vehicle's resting period. When performing a risk assessment based on the battery data and the ambient temperature data, and obtaining the corresponding risk assessment result, the following is included: The risk of abnormal charging and discharging of the battery is assessed based on the DC-DC output voltage, the DC-DC output current, and the battery voltage. Based on the static voltage data and the ambient temperature data, a preset regression algorithm is used to predict the natural power loss rate per unit time, and the risk of abnormal self-discharge of the battery is assessed based on the natural power loss rate. The power-off time of the battery and the number of times the power-on was triggered during the power-off period are obtained. Based on the power-off time, the number of power-on events, and the ambient temperature data, a preset reliability distribution model is used to establish a mapping relationship between the power-on frequency and the battery health status. Based on the mapping relationship, the current battery health status value is predicted to assess the capacity degradation risk of the battery. Time series data is acquired, and a preset time series analysis algorithm is used to predict the voltage change trend of the battery. When the predicted battery voltage is lower than a preset power-on failure threshold, the risk of startup failure is assessed. The time series data includes historical battery voltage data, historical ambient temperature data, and historical user behavior data. Obtain the actual battery natural loss rate and the actual battery health status value, determine the first residual between the actual battery natural loss rate and the predicted battery natural loss rate, determine the second residual between the actual battery health status and the predicted battery health status, and use a preset anomaly detection algorithm to judge the first residual and the second residual based on the first residual and the second residual, and assess potential fault risks based on the judgment results. Based on the aforementioned abnormal charging and discharging risks, abnormal self-discharge risks, capacity degradation risks, startup failure risks, and potential fault risks, a battery state risk assessment result is generated.

3. The method for predicting vehicle power-on faults according to claim 2, characterized in that, When assessing the risk of abnormal charging and discharging of the battery based on the DC-DC output voltage, the DC-DC output current, and the battery voltage, the following are included: The DC-DC output voltage and the DC-DC output current are compared with preset output voltage safety thresholds and preset output current safety thresholds, respectively, to obtain a first comparison result; The battery voltage is compared with a preset battery voltage threshold to obtain a second comparison result; The risk of abnormal charging and discharging of the battery is assessed based on the first comparison result and the second comparison result.

4. The method for predicting vehicle power-on faults according to claim 1, characterized in that, When performing a risk assessment based on the vehicle component data and calling the component evaluation model to obtain the corresponding risk assessment result, the following are included: Acquire the control signals and status feedback signals of the vehicle components; Based on the amplitude, frequency, and waveform characteristics of the control signal, assess whether there is a risk of control signal abnormality; Based on the correspondence between the control signal and the state feedback signal, the pull-in delay time is determined, and the risk of pull-in timing abnormality is assessed based on the pull-in delay time. Based on the abnormal risk of the control signal and the abnormal risk of the engagement timing, a risk assessment result of the component status is generated.

5. The method for predicting vehicle power-on faults according to claim 1, characterized in that, When performing a risk assessment based on the user behavior data and the ambient temperature data, and obtaining the corresponding risk assessment result by calling the user behavior risk assessment model, the process includes: Obtain historical user driving data; From the historical user behavior data and the ambient temperature data, vehicle usage intensity features, electricity usage intensity features, charging behavior features and environmental correlation features are extracted to construct a user behavior feature vector. Based on the user behavior feature vector, a preset learning algorithm is used to perform cluster analysis to divide users into multiple user groups with similar behavior patterns. Based on the historical vehicle power failure rate of each user group, a risk level and / or a risk score are assigned to each group. Based on the risk level and / or risk score, output the user behavior risk assessment result.

6. The method for predicting vehicle power-on faults according to claim 1, characterized in that, When predicting the risk level of a power-on failure in the vehicle based on the comprehensive risk assessment results, the following are included: When the comprehensive risk assessment result is mapped to a first numerical range, the predicted risk level of the vehicle experiencing a power-on failure is the first risk level. When the comprehensive risk assessment result is mapped to the second numerical range, the predicted risk level of the vehicle experiencing a power-on failure is the second risk level; When the comprehensive risk assessment result is mapped to the third numerical range, the predicted risk level of the vehicle experiencing a power-on failure is the third risk level. When the comprehensive risk assessment result is mapped to the fourth numerical range, the predicted risk level of the vehicle power-on failure is the fourth risk level, wherein the first risk level is higher than the second risk level, the second risk level is higher than the third risk level, the third risk level is higher than the fourth risk level, and the first numerical range to the fourth numerical range do not overlap and cover the full range of the comprehensive risk assessment result.

7. The method for predicting vehicle power-on faults according to claim 1, characterized in that, When performing a risk assessment based on the vehicle operation data and the ambient temperature data, and calling a preset assessment model corresponding to the vehicle component data, the battery data, and the user behavior data to obtain the corresponding risk assessment result, the process includes: The vehicle operation data is grouped according to the ambient temperature data and the user behavior data to form multiple verification datasets; The preset evaluation model is validated based on each validation dataset, and the vehicle operation data is continuously updated. The preset evaluation model is updated based on the verification results and the updated vehicle operation data.

8. The method for predicting vehicle power-on faults according to claim 6, characterized in that, After predicting the risk level of a power-on failure in the vehicle based on the comprehensive risk assessment results, the following steps are included: When the risk level is the first risk level, a first prompt message containing emergency alarm content is sent to the user; When the risk level is the second risk level, a second prompt message containing risk warning and usage optimization suggestions is sent to the user; When the risk level is the third or fourth risk level, an assessment report of normal vehicle status is generated for the user to query.

9. A predictive system for vehicle power-on faults, characterized in that, include: The data acquisition module is used to collect vehicle operation data and ambient temperature data. The vehicle operation data includes vehicle component data, battery data, and user behavior data. The assessment module is used to perform risk assessment based on the vehicle operation data and the ambient temperature data by calling a preset assessment model corresponding to the vehicle component data, the battery data and the user behavior data, and to obtain the corresponding risk assessment result. The preset assessment model includes a battery assessment model, a component assessment model and a user behavior risk assessment model. The generation module is used to perform weighted fusion analysis on the risk assessment results to generate a comprehensive risk assessment result; The prediction module is used to predict the risk level of the vehicle's power-on failure based on the comprehensive risk assessment results.

10. A vehicle, characterized in that, include: The vehicle power-on fault prediction system as described in claim 9; or, A processor, a memory, and a vehicle power-on fault prediction program stored in the memory and executable on the processor, wherein the vehicle power-on fault prediction program, when executed by the processor, implements the vehicle power-on fault prediction method as described in any one of claims 1-8.

Images