A vehicle-mounted battery multi-parameter edge early warning method and device based on habit modeling

Abstract

The application discloses a kind of based on habit modeling vehicle-mounted battery multi-parameter edge early warning method and device, it is related to automobile battery safety early warning technical field.This kind of based on habit modeling vehicle-mounted battery multi-parameter edge early warning method and device, including S1, battery state data, battery characteristic data and motion state data are collected and preprocessed, and standardization vehicle state data set is constructed;S2, the structural load state of battery is analyzed, and the edge side analysis link adjustment is driven;S3, the accumulation state of structural risk in battery is evaluated, and state analysis depth is corrected in real time;S4, the linkage degree of vehicle-mounted battery operation risk is evaluated, and the edge side hierarchical early warning strategy adjustment is driven;S5, battery operation is judged, and output is used guide and maintenance prompt.The problem that traditional early warning mode is difficult to perceive in time is solved in high-density commuting start-stop operating condition, and battery risk evolves in the form of accumulation hidden.

Description

Technical Field

[0001] This invention relates to the field of automotive battery safety early warning technology, specifically to a method and device for multi-parameter edge warning of vehicle batteries based on habit modeling. Background Technology

[0002] With the increasing density of urban traffic, on-board battery status monitoring technology has been widely applied in scenarios such as operational safety assurance, energy management, and lifespan assessment of new energy vehicles. Existing on-board battery monitoring solutions typically use voltage, current, and temperature sensors to periodically sample the operating status of individual battery cells and modules, and combine the sampled data to assess state of charge, health status, and instantaneous power output capability. To adapt to the non-steady-state operating characteristics of vehicles during acceleration, deceleration, and start-stop processes, some monitoring solutions also introduce processing logic such as sliding window statistics, simple filtering, or threshold holding to reduce the impact of transient disturbances on parameter interpretation, thereby providing a basic operational status basis for the vehicle control unit. In practical applications, the aforementioned monitoring data is often collected synchronously with vehicle speed, pedal opening, braking signals, and other operational information, and then fused and processed by edge computing units to support the execution of battery safety protection and energy allocation strategies.

[0003] For example, invention patent CN106680720B discloses a vehicle-mounted battery failure early warning system based on the Internet of Vehicles (IoV). The IoV OBD terminal acquires the voltage information of the vehicle-mounted battery and sends it to the IoV server. The IoV server stores the received voltage information and the time it was received. Based on a preset algorithm, the voltage information, and the time it was received, it determines whether the vehicle-mounted battery has failed. If the battery is determined to be failed, an early warning message is sent to the user terminal. This invention also discloses a vehicle-mounted battery failure early warning method based on the Internet of Vehicles.

[0004] For example, invention patent CN114280483B discloses a vehicle battery status monitoring system, method, and device. This system uses a monitoring device electrically connected to the vehicle battery and processor in an autonomous driving device to monitor the charging status of the vehicle battery. Based on the monitored charging status, the processor determines whether there is a short-circuit risk in the vehicle battery. If a short-circuit risk is identified, a warning is issued, and the autonomous driving device is controlled according to a pre-set degradation control strategy. This method adds a device to monitor the vehicle battery status between the vehicle battery and the processor in the autonomous driving device. When a short-circuit risk is identified, a degradation control strategy is used to control the autonomous driving device, preventing emergency braking caused by a short circuit.

[0005] However, in real-world urban commuting scenarios with frequent start-stop operations, vehicles exhibit characteristics of short-distance travel, multiple starts, and alternating low-speed, high-load operation. Batteries are constantly subjected to high-rate pulse discharge and repeated thermal expansion and contraction. Under these conditions, the fluctuations in instantaneous monitoring parameters such as cell voltage and surface temperature are often within normal ranges, making it difficult to directly reflect the evolutionary process of gradually increasing internal impedance and continuous concentration of localized thermal stress. Existing monitoring methods mostly rely on fixed thresholds or static models for judgment, lacking the ability to model the combined effects of driving start-stop habits, load rhythms, and behavioral patterns. This makes it impossible to identify the structural risk accumulation problems induced by high start-stop frequency and commuting usage characteristics, resulting in potential thermal runaway risks remaining hidden for a considerable period while parameters appear normal, indicating continuous structural deterioration.

[0006] To address the above issues, there is an urgent need for a method and device for multi-parameter edge warning of vehicle batteries based on habit modeling. Summary of the Invention

[0007] Technical problems to be solved

[0008] To address the shortcomings of existing technologies, this invention provides a method and device for multi-parameter edge warning of vehicle batteries based on habit modeling, which solves the problem that battery risks evolve covertly in a cumulative manner under high-density commuting start-stop conditions, making it difficult for traditional warning methods to detect them in a timely manner.

[0009] Technical solution

[0010] To achieve the above objectives, the present invention provides the following technical solution: a method and device for multi-parameter edge early warning of vehicle batteries based on habit modeling, comprising: S1, collecting battery status data, battery feature data, and motion state data during vehicle operation; preprocessing the battery status data, battery feature data, and motion state data to construct a standardized vehicle status dataset; S2, based on the standardized vehicle status dataset, performing load analysis on the battery structural load state from the battery pulse load evolution characteristics triggered by vehicle start-stop behavior, and driving dynamic adjustment of the edge-side analysis link based on the load analysis results; S3, based on the standardized vehicle status dataset, performing risk assessment on the cumulative state of internal structural risks of the battery, and correcting the depth of subsequent state analysis in real time based on the assessment results; S4, using the load analysis results and risk assessment results as input, performing linkage assessment on the degree of linkage of vehicle battery operation risks, and driving adjustment of edge-side hierarchical early warning strategy and service decision triggering based on the linkage assessment results; S5, continuously judging the battery operation status at the edge based on behavioral load changes and risk evolution state, and converting the judgment results into executable usage guidance and maintenance prompts before completing multi-terminal output and recording.

[0011] Further, the specific steps for collecting battery state data, battery characteristic data, and motion state data during vehicle operation, and preprocessing the battery state data, battery characteristic data, and motion state data to construct a standardized vehicle state dataset are as follows: Battery state data during vehicle operation is collected through a multi-parameter sensing interface module and sensor array deployed inside the battery. This battery state data includes: cell voltage, cell temperature, module temperature, battery pack current, and battery pack voltage at each sampling time. Battery characteristic data reflecting the internal structure and reaction state of the battery is collected through pressure sensors deployed in the battery pack exhaust channel and inside the casing, as well as a dedicated EIS interface. This battery characteristic data includes: internal pressure of the battery pack, concentration of characteristic gases, and high-frequency, mid-frequency, and low-frequency complex impedance values ​​in the electrochemical impedance spectrum generated by an impedance analyzer. Kinematic state data during vehicle operation is obtained by integrating a GPS and Beidou dual-mode positioning unit and a six-axis inertial navigation component. The data includes: the slope distribution of the road where the vehicle travels, the driving speed, and the change in the vehicle's steering angle; recording the number of sampling points included in each sampling time and within each sliding time window, and dividing a complete start, drive, braking, and idling process into a start-stop segment, recording the start and end times of each start-stop segment; employing multi-level noise reduction and outlier removal strategies for the collected battery status data, battery feature data, and motion status data: filtering high-frequency noise from the sensor using a sliding window filtering algorithm; removing outliers using the 3σ criterion; filling in missing data using a data completion algorithm based on linear interpolation; normalizing the standardized battery status data, battery feature data, and motion status data using extreme value linear normalization based on a sliding time window, and completing frame encapsulation in chronological order, with each frame containing a timestamp, parameter type identifier, value, acquisition precision identifier, and acquisition channel number; and writing the frames into a hierarchical buffer composed of eMMC and microSD in ascending order of timestamp between frames to construct a standardized vehicle status dataset.

[0012] Furthermore, the specific steps for analyzing the battery structural load state based on the battery pulse load evolution characteristics caused by vehicle start-stop behavior using a standardized vehicle state dataset are as follows: Calculate and sum the differences between the cell voltage value at the start of the start-stop segment and the cell voltage values ​​at each time point to obtain the cumulative voltage pulse value; use the difference between the maximum and minimum cell temperature values ​​in the start-stop segment as the numerator, and the difference between the end time and the start time of the start-stop segment plus one as the denominator to obtain the temperature change density; calculate the voltage pulse accumulation at each time point in the start-stop segment... The pressure hysteresis cumulative value is obtained by subtracting the internal pressure value of the battery pack at the start of the start-stop segment from the internal pressure value of the battery pack at the start of the start-stop segment. The difference is then summed. The gas hysteresis cumulative value is obtained by calculating the difference between the characteristic gas concentration value at each moment of the start-stop segment and the characteristic gas concentration value at the start of the start-stop segment. The gas hysteresis cumulative value is obtained by calculating the sum of the pressure hysteresis cumulative value and the gas hysteresis cumulative value, and then multiplying it by the voltage pulse cumulative value and the temperature change density. The multi-feature joint value is obtained by summing the multi-feature joint values ​​of all start-stop segments and dividing by the number of sampling points.

[0013] Furthermore, the specific steps for dynamically adjusting the edge-side analysis link based on load analysis results are as follows: On the edge side, using a sliding time window, the start-stop habit pulse load values ​​obtained in a continuous commuting scenario are organized into a time series to form a start-stop habit pulse load evolution sequence, and the growth rate, fluctuation range, and persistence characteristics of the start-stop habit pulse load evolution sequence on the time axis are recorded; the edge side compares the current start-stop habit pulse load value with the start-stop habit pulse load evolution sequence formed by the same vehicle within a historical time window, and calculates the relative position and offset trend of the start-stop habit pulse load value in the historical distribution. When the start-stop habitual pulse load value shows a monotonically increasing trend over N consecutive time windows, the edge side triggers adaptive enhancement measures for subsequent analysis processes, including increasing the impedance spectrum sampling trigger frequency, extending the statistical window length for temperature gradient and pressure changes, and retaining key data segments with higher time resolution in the local cache. Simultaneously, the start-stop habitual pulse load value and its offset trend are written into the edge side risk analysis context as behavioral inputs and associated with existing battery state data for storage. At the same time, the edge side outputs the offset trend of the start-stop habitual pulse load value to the vehicle and mobile terminals through a multi-protocol communication link.

[0014] Furthermore, the specific steps for risk assessment of the cumulative state of internal structural risks of the battery based on the standardized vehicle state dataset are as follows: Divide the peak-to-valley difference of the cell voltage within the start-stop segment by the maximum cell temperature value within the start-stop segment to obtain the start-stop cell load value; calculate the difference between the maximum and minimum cell temperature values ​​within the start-stop segment, and then divide this difference by the difference between the maximum and minimum module temperature values ​​within the start-stop segment to obtain the thermal distribution imbalance characterization value; calculate the ratio of the absolute value of the high-frequency complex impedance value to the absolute value of the mid-frequency complex impedance value in the impedance spectrum. Then, the absolute value of the comparison value is taken to obtain the impedance dissipation frequency band expansion value; the absolute value of the difference between the imaginary part of the low-frequency complex impedance value and the imaginary part of the mid-frequency complex impedance value is calculated, and then divided by the absolute value of the difference between the real part of the high-frequency complex impedance value and the real part of the mid-frequency complex impedance value plus one, to obtain the diffusion-dominant offset characterization value; the start-stop cell load value, the thermal distribution imbalance characterization value, the impedance dissipation frequency band expansion value and the diffusion-dominant offset characterization value are multiplied to obtain the structural coupling expansion intermediate value; the structural coupling expansion intermediate value is added by one and the natural logarithm is taken to obtain the structural risk accumulation characterization value.

[0015] Furthermore, the specific steps for real-time correction of subsequent state analysis depth based on evaluation results are as follows: After obtaining the cumulative characterization value of structural risk, the cumulative characterization value of structural risk is written into the edge-side operation state table as a structural state index, and a one-to-one mapping relationship is established with the electrochemical impedance spectroscopy, battery state data, and internal pressure value record of the battery pack corresponding to the current analysis window, forming a state index sequence with structural state as the main line; for the electrochemical impedance spectroscopy sampling results newly entering the analysis window, the edge side extracts the high-frequency complex impedance value, mid-frequency complex impedance value, and low-frequency complex impedance value respectively, and compares them segment by segment with the electrochemical impedance spectroscopy sampling results of the corresponding frequency bands in adjacent analysis windows. When each frequency band shows the same direction of shift within the same window and the shift amplitude remains continuously changing between adjacent windows, the current analysis window is marked as a significant structural change window, and the corresponding cell voltage change trajectory and internal pressure change of the battery pack within the significant structural change window are analyzed. Chemical records and temperature distribution data are independently sealed and stored. When the cumulative structural risk characterization value maintains a monotonically increasing characteristic within M consecutive windows, the electrochemical impedance spectroscopy analysis path is switched, and the conventional full-band scan results are split into high-frequency conductivity response segment, mid-frequency charge transfer segment, and low-frequency diffusion-restricted segment for separate storage. An index identifier corresponding to the cumulative structural risk characterization value is added to the stored record. In subsequent analysis, when a change in the morphology of the electrochemical impedance spectroscopy is detected, the corresponding historical structural evolution segment is retrieved first through the index of the cumulative structural risk characterization value. The start-stop habitual pulse load value, cell voltage value change trajectory, and internal pressure value change record of the battery pack under the same index are retrieved. The state evolution sequence before and after the structural change is aligned and compared to determine whether the structural change belongs to a stage in the continuous structural risk accumulation process. The determination result is recorded as a structural risk evolution path and written into the edge-side event log.

[0016] Furthermore, the specific steps for evaluating the linkage degree of vehicle battery operation risk using load analysis results and risk assessment results as input are as follows: Add one to the start-stop habitual pulse load value and take its natural logarithm to obtain the start-stop pulse load compression value; add one to the structural risk cumulative characterization value and take its natural logarithm to obtain the structural risk cumulative compression value; divide the battery pack current by the battery pack voltage to obtain the instantaneous electrical response characterization value; use the product of the start-stop pulse load compression characterization value, the structural risk cumulative compression value, and the instantaneous electrical response characterization value as the independent variable to perform arctangent mapping, and then multiply by 2π to obtain the edge-level linkage driving value.

[0017] Furthermore, the specific steps for adjusting the edge-side hierarchical early warning strategy and triggering service decisions based on the linkage evaluation results are as follows: Real-time comparison of the current edge-side hierarchical linkage driving value with the linkage driving threshold, where the linkage driving threshold includes a first driving threshold and a second driving threshold, and the first driving threshold is greater than the second driving threshold; when the edge-side hierarchical linkage driving value is less than or equal to the second driving threshold, the low-level warning status and current battery operating status data are sent to the vehicle instrument panel via the vehicle communication link, the corresponding low-risk warning indicator light illuminates on the vehicle instrument panel and remains in silent warning mode, and simultaneously, operating information containing a summary of the current battery status is pushed to the mobile terminal via the wireless communication link; battery status records and habit trajectory analysis results are periodically uploaded to the cloud side to form daily operating reports, and users are allowed to configure and adjust the sensor sampling frequency and data upload cycle through the mobile terminal and cloud interface; when the edge-side hierarchical linkage driving value is greater than the second driving threshold and less than or equal to the first driving threshold, a medium warning is output to the vehicle instrument panel via the vehicle communication link. The system generates a high-level warning signal, which illuminates an orange indicator light at the vehicle's endpoint and triggers an intermittent buzzer. Simultaneously, it pushes a warning message containing the warning level, risk source analysis information, and operational precautions to the mobile device via a wireless communication link. It also supports one-click sharing of warning information to the automaker's after-sales and rescue platforms via a message queue protocol. The system synchronously records warning logs and initiates risk trend analysis tasks on the cloud side, while providing interfaces for adjusting sampling frequency and warning-related parameters on both the mobile device and the cloud. When the edge-level linkage drive value exceeds the first drive threshold, a high-level warning signal is output to the vehicle's instrument panel via the vehicle communication link, illuminating a red indicator light at the vehicle's endpoint and triggering a continuous buzzer. Simultaneously, it links the vehicle's voice component to output safety prompts and pushes complete warning information, including the high-level warning status, risk source analysis results, and emergency response guidelines, to the mobile device via a wireless communication link. The system also uploads battery status data, warning event numbers, and analysis results to the cloud side in real time, and allows after-sales personnel to remotely configure the sampling frequency and analysis strategy via the cloud platform.

[0018] Furthermore, the specific steps for continuously judging the battery's operating status based on behavioral load changes and risk evolution at the edge, and converting the judgment results into executable usage guidance and maintenance prompts before completing multi-terminal output and recording are as follows: At the edge, acquire start-stop habitual pulse load values, structural risk accumulation characterization values, and edge-level linkage drive values, and simultaneously read electrochemical impedance spectroscopy acquisition results and vehicle motion state data to form a multi-source state input set for service decision-making; based on the morphological change characteristics of the electrochemical impedance spectroscopy, comprehensively analyze the current battery operating status to form a battery health status assessment result, and use the battery health status assessment result as the basic reference for subsequent service generation; combine the distribution characteristics of start-stop habitual pulse load values ​​in different time periods, battery temperature status, and battery health status assessment results... The system analyzes the matching degree between the user's existing charging behavior and the battery's carrying capacity, and generates charging period adjustment suggestions. During vehicle operation, based on the path slope distribution, driving speed, and real-time battery status changes, it analyzes the battery's electrical and thermal response status during the energy recovery phase. When a continuous downhill scenario is identified, corresponding driving and energy recovery guidance information is generated. When the battery health status assessment result shows a continuous downward trend, maintenance reminder information is generated, and the corresponding status basis and analysis results are output as maintenance reference content to prompt the user to perform battery testing and maintenance operations. The battery health assessment results, charging suggestions, driving guidance information, and maintenance reminders are synchronously output through the vehicle-to-mobile interaction link, and the corresponding service decision records and status data are written to cloud storage.

[0019] The second aspect of this invention provides a multi-parameter edge warning device for vehicle batteries based on habit modeling, comprising: a multi-parameter synchronous acquisition and packaging module for acquiring battery state data, battery feature data, and motion state data during vehicle operation, preprocessing the battery state data, battery feature data, and motion state data to construct a standardized vehicle state dataset; a pulse load characterization module for performing load analysis on the battery structural load state based on the standardized vehicle state dataset and the battery pulse load evolution characteristics caused by vehicle start-stop behavior, and driving dynamic adjustment of the edge-side analysis link based on the load analysis results; an impedance accumulation change analysis module for performing risk assessment on the accumulated state of internal structural risks of the battery based on the standardized vehicle state dataset, and real-time correction of the subsequent state analysis depth based on the assessment results; an edge hierarchical warning module for performing linkage assessment on the degree of linkage of vehicle battery operation risks using load analysis results and risk assessment results as input, and driving adjustment of edge-side hierarchical warning strategy and service decision triggering based on the linkage assessment results; and an application service linkage decision module for continuously judging the battery operation status at the edge based on behavioral load changes and risk evolution status, and converting the judgment results into executable usage guidance and maintenance prompts before completing multi-terminal output and recording.

[0020] Beneficial effects

[0021] The present invention has the following beneficial effects:

[0022] (1) The vehicle battery multi-parameter edge warning method and device based on habit modeling can continuously characterize the start-stop rhythm and load pulse behavior, so that structural changes inside the battery can be detected in advance before the parameters are abnormal, thereby improving the risk detection capability in commuting conditions.

[0023] (2) The vehicle battery multi-parameter edge warning method and device based on habit modeling, by classifying the intensity of state changes under behavior-driven, enables different risk stages to be matched with differentiated edge response strategies, thereby avoiding misjudgment caused by single threshold triggering.

[0024] (3) The vehicle battery multi-parameter edge warning method and device based on habit modeling combines start-stop behavior characteristics with thermoelectric response coupling relationship, so that risk judgment can reflect the load evolution law under real usage habits and enhance the adaptability of warning results to actual vehicle use scenarios.

[0025] (4) The vehicle battery multi-parameter edge warning method and device based on habit modeling directly maps the risk evolution results to the control and prompting actions on the edge side, thereby achieving close connection between the warning results and subsequent handling, and improving the timeliness and continuity of overall safety protection.

[0026] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0027] Figure 1 This is a flowchart of a multi-parameter edge warning method for vehicle batteries based on habit modeling according to the present invention.

[0028] Figure 2 This is a structural diagram of a vehicle-mounted battery multi-parameter edge warning device based on habit modeling according to the present invention;

[0029] Figure 3 This is a bar chart of edge hierarchical linkage driving values ​​under behavior-structure perturbation involved in the present invention.

[0030] Figure 4 This is a cross-sectional view of the internal multilayer PCB layout of the terminal box involved in this invention.

[0031] Figure 5 This is a circuit topology diagram of the power management module involved in this invention. Detailed Implementation

[0032] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0033] Please see Figures 1-5This invention provides a technical solution: a method and device for multi-parameter edge early warning of vehicle batteries based on habit modeling, comprising: S1, collecting battery status data, battery feature data, and motion state data during vehicle operation, preprocessing the battery status data, battery feature data, and motion state data to construct a standardized vehicle status dataset; S2, based on the standardized vehicle status dataset, performing load analysis on the battery structural load state from the battery pulse load evolution characteristics caused by vehicle start-stop behavior, and driving dynamic adjustment of the edge-side analysis link based on the load analysis results; S3, based on the standardized vehicle status dataset, performing risk assessment on the cumulative state of internal structural risks of the battery, and correcting the depth of subsequent state analysis in real time based on the assessment results; S4, using the load analysis results and risk assessment results as input, performing linkage assessment on the degree of linkage of vehicle battery operation risks, and driving adjustment of edge-side hierarchical early warning strategy and service decision triggering based on the linkage assessment results; S5, continuously judging the battery operation status on the edge side based on behavioral load changes and risk evolution state, and converting the judgment results into executable usage guidance and maintenance prompts before completing multi-terminal output and recording.

[0034] Specifically, the process involves collecting battery state data, battery characteristic data, and motion state data during vehicle operation. Preprocessing this data to construct a standardized vehicle state dataset involves the following steps: Continuously collecting battery state data during vehicle operation using a multi-parameter sensing interface module and sensor array deployed inside the battery. This collection process synchronously acquires cell voltage, cell temperature, module temperature, battery pack current, and battery pack voltage at each sampling moment using a unified time reference, thus forming a basic state data set reflecting the evolution of the battery's electrical and thermal states. Simultaneously, pressure sensors deployed in the battery pack's exhaust channels and inside the casing, combined with battery... The reserved EIS dedicated interface synchronously acquires battery characteristic data reflecting changes in the battery's internal structure and reaction characteristics. This includes the internal pressure value of the battery pack, the concentration of characteristic gases, and electrochemical impedance spectroscopy data output by the impedance analyzer. High-frequency complex impedance values, mid-frequency complex impedance values, and low-frequency complex impedance values ​​are extracted from the impedance spectrum to characterize changes in conductive paths, charge transfer, and diffusion-limited processes. Furthermore, by integrating a GPS and BeiDou dual-mode positioning unit and a six-axis inertial navigation component, the kinematic state of the vehicle during operation is collected, acquiring the path gradient distribution, driving speed, and steering angle changes during driving, enabling the battery operating state to establish a correspondence with specific driving conditions.

[0035] At the data organization level, the number of sampling points included in each sampling time and sliding time window is recorded. The start-stop segment is divided according to a complete start, drive, brake and idle process of the vehicle, and the start and end times of each start-stop segment are clearly defined to ensure that subsequent analysis can be carried out with start-stop behavior as the basic time unit. Subsequently, multi-level noise reduction and anomaly removal processing are performed on the collected battery status data, battery feature data and motion status data. High-frequency noise introduced during sensor sampling is reduced by sliding window filtering. Data points that deviate significantly from the normal distribution are removed based on the three-standard-deviation criterion. The missing sampling points are filled in by linear interpolation, thereby ensuring the continuity and consistency of the data sequence in the time dimension.

[0036] After noise reduction, anomaly removal, and completion processing, the standardized battery state data, battery feature data, and motion state data undergo extreme value linear normalization processing based on a sliding time window to make different physical quantities comparable on a numerical scale. The data is then framed and encapsulated in chronological order, with each data frame containing a timestamp, parameter type identifier, numerical information, acquisition precision identifier, and acquisition channel number. Each data frame is written to a hierarchical cache consisting of eMMC and microSD in ascending order of timestamp, thereby building a standardized vehicle state dataset on the vehicle side that is structurally consistent, time-aligned, and traceable. This provides a reliable data foundation for subsequent start-stop habit modeling, structural risk analysis, and edge linkage decision-making.

[0037] In this implementation plan, a multi-source state data foundation that is time-aligned, physically consistent, and traceable is built on the vehicle side, providing reliable data input conditions for subsequent start-stop habit modeling, structural risk accumulation characterization, and edge hierarchical linkage decision-making. By synchronously collecting and uniformly organizing battery electrical state, thermal state, internal structural response, and vehicle kinematic state, data from different sources and with different dimensions form a complete mapping relationship on the same time axis, thus avoiding the mismatch between behavior and state caused by asynchronous sampling and data loss. Through start-stop segmentation and sliding time window organization, battery state changes can be matched one-to-one with specific vehicle usage behaviors, ensuring that subsequent analysis is carried out based on the actual operating process as the basic unit. Through multi-level noise reduction, anomaly removal, and data completion processing, the interference of sensor noise and occasional anomalies on the analysis results is reduced, improving the reliability of state data in terms of continuity and stability. Through normalization and frame encapsulation based on sliding time windows, multi-parameter data is transformed into a standardized dataset with a unified structure, comparable scale, and direct use for calculation and storage, thus meeting the technical prerequisite for completing the entire process of "collection-processing-analysis-decision" at the edge, laying the foundation for subsequent risk identification and linkage output.

[0038] Specifically, based on a standardized vehicle state dataset, the load analysis of the battery structural load state based on the battery pulse load evolution characteristics caused by vehicle start-stop behavior involves the following steps: Calculate and sum the differences between the cell voltage value at the start of the start-stop segment and the cell voltage values ​​at each other time point to obtain the cumulative voltage pulse value; use the difference between the maximum and minimum cell temperature values ​​in the start-stop segment as the numerator, and the difference between the end time and the start time of the start-stop segment plus one as the denominator to obtain the temperature change density; calculate the voltage pulse intensity at each time point in the start-stop segment... The pressure hysteresis cumulative value is obtained by subtracting the internal pressure value of the battery pack from the internal pressure value of the battery pack at the start of the start-stop segment and summing the results. The gas hysteresis cumulative value is obtained by calculating the difference between the characteristic gas concentration value at each moment in the start-stop segment and the characteristic gas concentration value at the start of the start-stop segment. The gas hysteresis cumulative value is obtained by calculating the sum of the pressure hysteresis cumulative value and the gas hysteresis cumulative value, and then multiplying it by the voltage pulse cumulative value and the temperature change density. The multi-feature joint value is obtained by adding the multi-feature joint values ​​of all start-stop segments and dividing by the number of sampling points.

[0039] The formula for calculating the start / stop habitual pulse load value is as follows:

[0040] ;

[0041] In the formula, Represents the set of start / stop segments used to constrain the process. The sliding time window is used to limit the start and stop segments involved in the calculation to only those from the current analysis window, thereby ensuring that the start and stop habitual pulse load values ​​reflect the recent operating status rather than historical cumulative results. It represents the current sampling time in the unified timestamp sequence, used to identify the synchronization position of multi-parameter data on the edge side, ensuring the consistency of data from different sensing channels in the time dimension, and originates from the millisecond-level timestamp sequence generated by GPS timing in the terminal box; This indicates the number of sampling points included within the sliding time window, used to limit the time range of start-stop behavior and pulse load statistics, so that the analysis results reflect the overall characteristics within a continuous commuting cycle, and are derived from the windowed segmentation processing results of the timestamp sequence in the terminal box; This represents the i-th start-stop segment, used to describe a complete start-up, driving, braking, and idling process. It is the basic analysis unit for subsequent pulse load and thermal stress calculations, derived from the judgment results of the changes in acceleration sign and velocity in adjacent sampling points. It represents the start time of the i-th start-stop segment, used to determine the baseline state of each parameter at the start of the start-stop behavior, and is derived from the timestamp positions corresponding to the change of acceleration sign and the entry of velocity from zero and low speed into the change range; Indicates the end time of the i-th start-stop segment, used to define the termination position of the influence range of the start-stop behavior, derived from the timestamp position corresponding to the change of acceleration sign again and the velocity entering the stable range; This represents the cell voltage value at the start of the i-th start-stop segment, which serves as a reference for voltage changes within that segment. It is derived from the sampling results of the voltage acquisition channel at the start timestamp of the start-stop segment. This represents the cell voltage value at sampling time j, which reflects the voltage response state of the battery under pulse discharge conditions during the start-stop process. It is derived from the real-time acquisition results of cell and module voltages by the voltage acquisition channel in the multi-parameter sensing interface module. This represents the cell temperature value at sampling time j, which reflects the thermal response state of the cell under start-stop pulse load conditions. It is derived from real-time temperature data collected by the cell-level temperature sensor. The internal pressure value of the battery pack at sampling time j is used to reflect the pressure response caused by the generation and volume change of gas inside the battery under start-stop load conditions. It is derived from the pressure sensor installed at the exhaust port of the battery pack. This represents the internal pressure value of the battery pack at the start time of the i-th start-stop segment, used to calculate the cumulative pressure change within that segment. It is derived from the sampling results of the pressure sensor at the start timestamp of the start-stop segment. This represents the characteristic gas concentration value at sampling time j, used to reflect the gas release trend of the internal side reaction of the battery under start-stop pulse load conditions, and is derived from the real-time detection results of the characteristic gas by the gas sensor array. The characteristic gas concentration at the start time of the i-th start-stop segment is used to measure the relative cumulative change in gas concentration during the start-stop behavior, and is derived from the sampling results of the gas sensor at the start timestamp of the start-stop segment.

[0042] This implementation scheme constructs a continuous behavioral index that reflects the accumulated state of structural load by uniformly quantifying the intrinsic correlation between frequent vehicle start-stop behavior and the multi-physical response of the battery. This index is used to reveal the hidden precursors of thermal runaway in high-density urban commuting scenarios, where "parameters appear normal on the surface, but internal risks continue to accumulate." Specifically, the formula uses start-stop segments as the basic analysis unit, jointly calculating the pulse discharge, battery thermal expansion and contraction changes, and the hysteretic cumulative response of pressure and characteristic gases generated during start-stop processes. This amplifies and explicitly expresses the repetitive pulse stress caused by high start-stop frequency and low-speed, high-load driving. By statistically analyzing multiple start-stop segments through a sliding time window, the formula can eliminate the influence of occasional fluctuations in a single start-stop event, highlighting the long-term trend of behavioral patterns on the internal state of the battery within the commuting cycle. The calculation results of this formula reflect the coupling relationship between the intensity of start-stop behavior and the degree of battery response in a continuous numerical form, providing a unified behavioral input basis for subsequent edge-side analysis. It can detect the formation process of internal impedance evolution and local thermal stress concentration risks in advance before obvious abnormalities in temperature and voltage appear, thereby supporting the connection between trend analysis, data link strengthening and risk analysis under threshold-free conditions, and forming an important foundation for early warning of vehicle batteries based on driving habits.

[0043] Specifically, the dynamic adjustment of the edge-side analysis link driven by load analysis results involves the following steps: The edge side is handled by a terminal box deployed inside the vehicle. This terminal box integrates data acquisition, caching, computing, and communication capabilities, acting as an in-vehicle edge computing entity to perform start-stop habit analysis and risk assessment functions. Within this edge-side terminal box, a sliding time window method is used to organize the start-stop habit pulse load values ​​obtained in continuous commuting scenarios into a time-series format, forming a chronologically ordered evolution sequence of start-stop habit pulse load. Within each sliding time window, the amplitude of change, fluctuation range, and duration of the corresponding start-stop habit pulse load value on the time axis are recorded to characterize the phased evolution characteristics of start-stop behavior intensity.

[0044] The edge-side terminal box further compares and analyzes the start-stop habit pulse load value obtained at the current moment with the start-stop habit pulse load evolution sequence formed by the same vehicle in the past operation process. The historical time window corresponding to the past operation process is a set of multiple sliding time windows continuously recorded by the terminal box in normal operation. By comparing the relative position of the current start-stop habit pulse load value in the evolution sequence, the offset direction and offset magnitude relative to the historical distribution are determined, which is used to reflect whether the current vehicle use behavior shows a continuous strengthening trend.

[0045] When the edge terminal box detects that the start-stop habitual pulse load value shows a window-by-window increasing change characteristic within N consecutive sliding time windows, it is determined that the intensity of the start-stop behavior has formed a continuous superposition state within the current commuting cycle. Here, N is used to characterize the number of sliding time windows representing the temporal continuity of the start-stop behavior. This number is automatically determined by the edge terminal box based on the duration of the sliding time windows and the distribution density of start-stop events within the corresponding time range. This is used to ensure that the window-by-window increasing characteristic reflects the continuous accumulation process of the behavior intensity rather than an instantaneous concentration phenomenon. The edge terminal box then triggers adaptive enhancement processing measures for subsequent analysis processes, including increasing the sampling trigger density of electrochemical impedance spectroscopy, extending the statistical time range of temperature gradients at the cell level and module level and internal pressure changes in the battery pack, and retaining higher time resolution key voltage and temperature data fragments in the local cache to enhance the ability to capture evidence of structural changes.

[0046] Meanwhile, the start-stop habitual pulse load value and its offset trend in the historical evolution sequence are written as behavioral state variables into the risk analysis context of the edge terminal box. Together with the battery status and wireless communication link in the corresponding time period, the offset trend of the start-stop habitual pulse load value and the corresponding behavioral change information are synchronously output to the vehicle display unit and mobile application to present the user with the correlation between the current vehicle use behavior and the battery load status.

[0047] In this implementation plan, a continuous monitoring and adaptive control mechanism based on the evolution of start-stop behavior intensity is established at the vehicle edge. This mechanism can promptly identify the continuous pulse load superposition trend of high-frequency start-stop behavior on the battery without relying on fixed threshold judgment. By organizing and serializing the start-stop habitual pulse load values ​​using a sliding time window, scattered instantaneous behavioral responses are transformed into evolutionary characteristics with temporal continuity, thus distinguishing between occasional start-stop fluctuations and long-term behavioral reinforcement processes. By comparing the current start-stop habitual pulse load values ​​with historical evolution sequences, the direction and degree of their offset in their own distribution can be quantified, enabling the determination of whether current vehicle usage behavior is gradually deviating from the existing operating state. When the start-stop habitual pulse load value is detected to be continuously rising within multiple consecutive time windows, adaptive reinforcement measures for subsequent analysis processes are triggered, ensuring that the accuracy of the collection and analysis of impedance, temperature, and pressure structural evidence increases synchronously with the increase in behavioral risk, thereby avoiding monitoring blind spots in the stage where behavioral load has significantly increased but structural risk has not yet become apparent. Simultaneously, the start-stop habitual pulse load values ​​and their offset trends are written into the edge-side risk analysis context and output externally, providing a clear behavioral-side reference basis for subsequent structural risk analysis and linkage decision-making, enhancing the interpretability and foresight of the overall early warning and service decision-making process.

[0048] Specifically, based on a standardized vehicle state dataset, the risk assessment steps for the cumulative state of internal structural risks in the battery are as follows:

[0049] Dividing the peak-to-valley difference of the cell voltage within the start-stop segment by the maximum cell temperature within the start-stop segment yields the start-stop cell load value. This value characterizes the relationship between the intensity of electrical fluctuations and the instantaneous thermal state exhibited by the cell under the action of pulse current during a complete start-stop process. The voltage peak-to-valley difference reflects the degree of electrical load impact at the moment of start-stop, while the maximum cell temperature reflects the thermal bearing level of this impact. The ratio of the two can characterize the degree of electrical load concentration under a unit thermal state.

[0050] The difference between the maximum and minimum cell temperature values ​​in the start-stop segment is calculated, and then divided by the difference between the maximum and minimum module temperature values ​​in the start-stop segment to obtain the thermal distribution imbalance characterization value. This value is used to reflect the degree of deviation of the temperature distribution inside the cell from the overall temperature distribution of the module. The cell temperature difference reflects local heat concentration and hot spot phenomena, while the module temperature difference reflects the temperature distribution range under overall heat dissipation conditions. The ratio of the two can reveal the imbalance of heat at the micro level relative to the macro level.

[0051] The ratio of the absolute value of the high-frequency complex impedance to the absolute value of the mid-frequency complex impedance in the impedance spectrum is calculated, and the absolute value of the ratio is taken to obtain the impedance dissipation frequency band expansion value. This value is used to characterize the impedance distribution relationship between the battery in the conductivity response-dominant region and the charge transfer-dominant region. The high-frequency complex impedance mainly corresponds to ohmic conduction and the fast interface response process, while the mid-frequency impedance mainly corresponds to the charge transfer and reaction kinetic process. This ratio reflects the degree of expansion of the dissipation-related impedance in different frequency bands.

[0052] The absolute value of the difference between the imaginary part of the low-frequency complex impedance and the imaginary part of the mid-frequency complex impedance is calculated, and then divided by the absolute value of the difference between the real part of the high-frequency complex impedance and the real part of the mid-frequency complex impedance plus one, to obtain the diffusion-dominant shift characterization value. This value is used to characterize the shift intensity of the diffusion-limited process relative to the conduction and charge transfer processes. The low-frequency imaginary part mainly reflects the characteristics of material diffusion and concentration polarization, the mid-frequency imaginary part reflects the reaction-related hysteresis characteristics, and the difference between the high-frequency and mid-frequency real parts reflects the change in the conduction path. This construction can highlight the relative shift degree of diffusion-dominant behavior in the overall impedance structure, and at the same time, the addition operation avoids numerical instability caused by the denominator approaching zero.

[0053] Multiplying the start-stop cell load value, thermal distribution imbalance characterization value, impedance dissipation frequency band expansion value, and diffusion-dominant offset characterization value yields the intermediate value of structural coupling expansion. This product form is used to enhance the coupling amplification effect when multiple physical mechanisms are simultaneously abnormal, so that when the electrical load, thermal imbalance, dissipation characteristics, and diffusion offset are significantly enhanced in any dimension, the overall structural response can be amplified synchronously, thereby improving the sensitivity to multi-mechanism coordinated anomalies, while preventing over-response when a single mechanism fluctuates slightly.

[0054] The cumulative structural risk characterization value is obtained by adding one to the intermediate value of the structural coupling expansion and taking the natural logarithm. The logarithmic compression is used to suppress the direct amplification effect of the differences in the dimensions of different physical quantities and extreme values ​​on the results, so that the structural risk exhibits a smooth increasing characteristic as the multi-mechanism coupling is enhanced, while ensuring sufficient resolution in the low-risk range. This is more conducive to the stable characterization and comparative analysis of the continuous evolution trend of structural risk at the edge.

[0055] The formula for calculating the cumulative characteristic value of structural risk is as follows:

[0056] ;

[0057] In the formula, This indicates the time index corresponding to the current analysis window. It is used to identify the correspondence of multi-channel acquired data under a unified time base. It is the basic identifier for realizing synchronous analysis of impedance change and thermal state. It comes from the timestamp sequence generated by the edge data processing unit. It represents the peak-to-valley difference of cell voltage within the start-stop segment, used to quantify the pulse electrical load amplitude borne by the battery during a single or multiple start-stop processes. It is a basic electrical indicator for characterizing the intensity of start-stop pulses and is derived from the continuous sampling data of the cell voltage acquisition channel within the start-stop segment. It represents the cell temperature value, which reflects the thermal response state of a single cell under start-stop load conditions. It is a basic thermal index that characterizes local thermal state changes and is derived from the data collected by temperature sensors deployed on the cell surface. The module temperature value reflects the overall thermal distribution of the battery module and is a basic thermal indicator for characterizing the macroscopic thermal diffusion level. It is derived from the temperature acquisition channels located at the battery module. It represents the high-frequency complex impedance value in the impedance spectrum, which is used to reflect the ohmic response characteristics of the battery under high-frequency excitation conditions. It is a basic impedance index characterizing the state of the conductive path and comes from the high-frequency sampling results when the impedance spectrum is scanned by the EIS dedicated interface. It represents the mid-frequency complex impedance value in the impedance spectrum, which is used to reflect the charge transfer-related response state of the battery in the mid-frequency range. It is the core reference quantity for describing the arc structure of the impedance spectrum and is derived from the mid-frequency sampling results obtained by EIS impedance spectrum scanning. The imaginary part of the low-frequency complex impedance value is used to reflect the impedance components related to the diffusion process and polarization behavior. It is a key indicator for characterizing the changes in diffusion-limited characteristics and is derived from the analytical results of the imaginary part of the low-frequency impedance sampling value. The imaginary part of the mid-frequency complex impedance value is used as a mid-frequency reference for comparing low-frequency diffusion characteristics. It is an important indicator for analyzing the degree of impedance spectrum morphology expansion and is derived from the analytical results of the imaginary part of the mid-frequency impedance sampling value. The real part of the high-frequency complex impedance value is used to reflect the impedance components related to resistive dissipation. It is a basic indicator for judging changes in the state of the conductive path and is derived from the analytical results of the real part of the high-frequency impedance sampling value. The real part of the intermediate frequency complex impedance value is used as an intermediate frequency reference for comparing impedance spectrum morphology. It is a fundamental indicator for analyzing the difference between ohmic response and charge transfer response, and is derived from the analytical results of the real part of the intermediate frequency impedance sampling value.

[0058] In this implementation scheme, the inherent coupling relationship between the frequency domain morphology changes of the battery impedance spectrum under start-stop pulse load and the imbalance state of thermal distribution between the cell and the module is uniformly quantified. This allows the structural risks corresponding to the cumulative impedance changes and local thermal stress concentrations to be explicitly expressed in continuous numerical form, thereby transforming the hidden evolution of internal battery risks into an analyzable state. The formula characterizes the pulse load intensity through the voltage peak-to-valley difference within the start-stop segment, the heat concentration trend on a spatial scale through the ratio of the temperature range between the cell and the module, the change in dissipation characteristics related to the conductive path through the ratio of high-frequency to mid-frequency impedance amplitudes, the enhancement of the diffusion-dominated response through the relative shift between the low-frequency imaginary part and the high-frequency real part, and the suppression of single-wave amplification under nonlinear compression mapping to highlight the continuous evolution characteristics, enabling the impedance spectrum morphology changes and thermal stress concentrations to be synergistically reflected within the same computational framework. The cumulative structural risk characterization value can continuously reflect the evolution trend of the internal structural state of the battery under operating conditions where temperature and voltage transient changes are not significant. This provides a stable and interpretable state-side quantitative basis for subsequent analysis of link regulation impedance changes and local thermal stress concentration analysis under threshold-free conditions.

[0059] Specifically, the steps for real-time correction of subsequent state analysis depth based on evaluation results are as follows: After obtaining the cumulative characterization value of structural risk, the cumulative characterization value of structural risk is written into the edge-side operation state table as a structural state index, and a one-to-one mapping relationship is established with the electrochemical impedance spectroscopy, battery state data, and internal pressure value record of the battery pack corresponding to the current analysis window, forming a state index sequence with structural state as the main line, which is used to continuously track the evolution process of the internal structure of the battery in the time dimension; For the electrochemical impedance spectroscopy sampling results newly entering the analysis window, the edge side extracts the high-frequency complex impedance value, mid-frequency complex impedance value, and low-frequency complex impedance value respectively, and compares them segment by segment with the electrochemical impedance spectroscopy sampling results of the corresponding frequency bands in adjacent analysis windows. When each frequency band shows the same direction shift in the same analysis window, and the shift maintains the same direction and shows a continuous changing trend between adjacent analysis windows, the current analysis window is marked as a significant structural change window, and the cell voltage change trajectory, internal pressure change record of the battery pack, and temperature distribution data corresponding to the significant structural change window are independently sealed and stored.

[0060] When the cumulative structural risk characterization value maintains a window-by-window increasing characteristic within M consecutive adjacent analysis windows, the structural risk is determined to be in a state of continuous accumulation. Here, M represents the number of analysis windows used to characterize the temporal continuity of the structural risk. This number is automatically determined by the edge side based on the duration of the analysis window and the sampling period of the electrochemical impedance spectroscopy to ensure that the increasing characteristic has temporal stability rather than being caused by a single perturbation. After the above conditions are met, the edge side switches the electrochemical impedance spectroscopy analysis path, splits the conventional full-band scan results into a high-frequency conductivity response segment, a mid-frequency charge transfer segment, and a low-frequency diffusion-restricted segment for separate storage, and adds an index identifier corresponding to the cumulative structural risk characterization value to the corresponding storage record.

[0061] In subsequent analysis, when a change in the electrochemical impedance spectroscopy morphology is detected, the edge side first retrieves the corresponding historical structural evolution fragments through the structural risk accumulation characterization value index, retrieves the start-stop habitual pulse load values, cell voltage value change trajectories, and battery pack internal pressure value change records recorded under the same index, and performs time-aligned comparison of the multi-source state evolution sequence before and after the structural change to determine whether the structural change belongs to a stage in the continuous structural risk accumulation process, and writes the determination result as a structural risk evolution path record into the edge side event log.

[0062] In this implementation plan, a traceable analysis mechanism based on the accumulation of structural risks is established at the edge, enabling continuous identification, location, and tracing of internal structural changes in the battery over time. By using the accumulated structural risk characterization value as a unified structural state index and mapping it one-to-one with electrochemical impedance spectroscopy, battery electrical state, and internal pressure records, a clear correspondence is established between different physical evidence within the same analysis window, thus avoiding the difficulty in corroborating structural changes scattered across multiple data sources. By discriminating between the unidirectional shift and continuous changes of high-frequency, mid-frequency, and low-frequency impedance characteristics within adjacent windows, random measurement fluctuations and the actual structural evolution process can be distinguished. Representative structural change windows are individually marked and archived to ensure data integrity during key evolution stages. When the accumulated structural risk characterization value shows a continuous increase... The system automatically adjusts the impedance spectrum analysis and storage strategy, shifting the data organization mode from conventional monitoring to structural evolution tracking, thereby improving the ability to analyze and distinguish changes in conduction paths, restricted charge transfer, and diffusion obstruction mechanisms. Furthermore, through historical retrieval based on structural state index and multi-source state alignment comparison, newly emerging impedance spectrum morphological changes can be quickly placed into the existing structural evolution trajectory for judgment, thereby identifying whether they belong to a stage in the continuous accumulation of structural risks. The judgment result is recorded in the form of events, providing a clear and interpretable structural evidence chain for subsequent risk evolution analysis, coordinated decision-making, and service output.

[0063] Specifically, using load analysis results and risk assessment results as input, the specific steps for evaluating the linkage degree of vehicle battery operation risk are as follows: add one to the start-stop habitual pulse load value and take the natural logarithm to obtain the start-stop pulse load compression value; add one to the structural risk cumulative characterization value and take the natural logarithm to obtain the structural risk cumulative compression value; divide the battery pack current by the battery pack voltage to obtain the instantaneous electrical response characterization value; use the product of the start-stop pulse load compression characterization value, the structural risk cumulative compression value, and the instantaneous electrical response characterization value as the independent variable to perform arctangent mapping, and then multiply by 2π to obtain the edge-level linkage driving value.

[0064] The formula for calculating the edge-level linkage driving value is:

[0065] ;

[0066] In the formula, It represents the pulse load value of start-stop habits, which is used to quantify the cumulative state of pulse electrical load on the battery under the condition of frequent start-stop of the vehicle. It is a basic indicator reflecting the degree of impact of vehicle use behavior on the battery. It represents the cumulative structural risk characterization value, used to quantify the evolution state of internal structural risk corresponding to the change in impedance spectrum morphology and the imbalance of thermal distribution in the cell module. It is a basic indicator reflecting the degree of change in the internal state of the battery. The current of the battery pack is used to reflect the instantaneous energy throughput intensity of the battery in actual operation. It is a basic electrical indicator for characterizing the real-time load level and is derived from the real-time measurement results of the battery pack current by the current acquisition channel. The current battery pack voltage reflects the instantaneous electrical energy state of the battery during actual operation. It is a basic electrical indicator for characterizing the power supply level and is derived from the real-time measurement results of the battery pack voltage by the voltage acquisition channel.

[0067] In this implementation example, the start / stop habitual pulse load value of running window 1 is set to 0.60, the structural risk cumulative characterization value is set to 0.50, the battery pack current is set to 42, and the battery pack voltage is set to 360.

[0068] Set the start / stop habit pulse load value of running window 2 to 0.95, the structural risk cumulative characterization value to 0.85, the battery pack current to 48, and the battery pack voltage to 357.

[0069] Set the start / stop habit pulse load value of running window 3 to 1.40, the structural risk cumulative characterization value to 1.10, the battery pack current to 55, and the battery pack voltage to 354.

[0070] Set the start / stop habit pulse load value of running window 4 to 1.20, the structural risk cumulative characterization value to 1.00, the battery pack current to 50, and the battery pack voltage to 355.

[0071] Set the start / stop habit pulse load value of running window 5 to 1.80, the structural risk cumulative characterization value to 1.60, the battery pack current to 58, and the battery pack voltage to 350.

[0072] Set the start / stop habit pulse load value of running window 6 to 2.30, the structural risk cumulative characterization value to 2.00, the battery pack current to 62, and the battery pack voltage to 348.

[0073] Set the start / stop habit pulse load value of running window 7 to 2.10, the structural risk cumulative characterization value to 1.90, the battery pack current to 57, and the battery pack voltage to 349.

[0074] Set the start / stop habit pulse load value for operating window 8 to 2.80, the structural risk cumulative characterization value to 2.50, the battery pack current to 65, and the battery pack voltage to 345. Calculate the edge-level linkage drive values ​​for each operating window, as shown in Table 1.

[0075] Table 1. Edge-level linkage driving value data table

[0076] Run window number Start-stop habitual pulse load value Cumulative Characteristic Value of Structural Risk Battery pack current Battery pack voltage Edge-level linkage driving value 1 0.60 0.50 42 360 0.057 2 0.95 0.85 48 357 0.089 3 1.40 1.10 55 354 0.128 4 1.20 1.00 50 355 0.108 5 1.80 1.60 58 350 0.171 6 2.30 2.00 62 348 0.206 7 2.10 1.90 57 349 0.184 8 2.80 2.50 65 345 0.244

[0077] like Figure 3 As shown in Table 1, this is a bar chart of edge hierarchical linkage driving values ​​under behavior-structure perturbation provided in this application example. Figure 3As can be seen from the evolution results of operating windows 1 to 8, the edge-level linkage driving value generally shows a trend of gradually increasing with the increase of both behavioral disturbances and structural risks. Among them, the edge-level linkage driving value corresponding to operating window 8 is the highest, indicating that the start-stop habitual pulse load level and the structural risk accumulation state are both in a high range during this operating stage. The comprehensive linkage pressure borne by the battery under actual working conditions is significantly enhanced. This window is more suitable as the key focus area for edge-side linkage response and service intervention. In contrast, the edge-level linkage driving value of operating window 1 is the lowest, indicating that the intensity of start-stop behavior is low and structural risks have not yet accumulated significantly during this stage. It will automatically maintain the basic monitoring state without triggering the enhanced linkage strategy, thereby avoiding unnecessary resource occupation. The bar chart of edge-level linkage driving values ​​under behavior-structure disturbances intuitively reflects the differences in edge-level linkage responses under different combinations of behavior intensity and structural risk. The higher the edge-level linkage driving value, the stronger the coupling between behavior disturbance and structural risk, and the more suitable it is to trigger subsequent linkage analysis, service decision-making, and risk warning. On the other hand, the operating window with the driving value in the lower range automatically reduces the priority of linkage response, reflecting the effective adjustment capability of this indicator to the linkage decision-making rhythm in complex operating scenarios.

[0078] Specifically, the steps for driving the adjustment of the edge-side hierarchical early warning strategy and triggering service decisions based on the linkage evaluation results are as follows: compare the current edge hierarchical linkage driving value with the linkage driving threshold in real time. The linkage driving threshold includes a first driving threshold and a second driving threshold, wherein the first driving threshold is greater than the second driving threshold.

[0079] When the edge-level linkage drive value is less than or equal to the second drive threshold, the low-level warning status and current battery operating status data are sent to the vehicle instrument panel via the vehicle communication link. The vehicle communication link uses a CAN bus-based vehicle communication protocol to complete data transmission. The corresponding low-risk warning indicator light on the vehicle instrument panel illuminates and remains in silent warning mode. At the same time, the operation information containing a summary of the current battery status is pushed to the mobile terminal via the wireless communication link. The wireless communication link uses cellular network and vehicle WiFi network for data transmission. The mobile terminal and the edge side receive the operation information through message publishing based on the MQTT protocol. On the cloud side, the edge side uploads battery status records and habit trajectory analysis results via the cellular communication network according to a preset period. The cloud summarizes the uploaded data to form a daily operation report and provides users with access to configure and adjust the sensor sampling frequency and data upload period through the mobile application interface and the cloud management interface.

[0080] When the edge-level linkage drive value is greater than the second drive threshold and less than or equal to the first drive threshold, a medium-level warning signal is output to the vehicle instrument through the vehicle communication link. The vehicle communication link completes the real-time transmission of the warning signal based on the CAN bus, and an orange indicator light illuminates at the vehicle end and triggers an intermittent buzzer. At the same time, a warning message containing the warning level, risk source analysis information, and operation precautions is pushed to the mobile terminal through the wireless communication link. The mobile terminal warning message is transmitted through a real-time push mechanism based on the MQTT or WebSocket protocol, and supports one-click forwarding of the warning information to the vehicle manufacturer's after-sales platform and rescue platform through an interface based on the message queue protocol. The corresponding warning log is recorded synchronously on the cloud side and a risk trend analysis task is started. At the same time, the online adjustment function of sensor sampling frequency and warning-related analysis parameters is opened through the cloud management interface and the mobile terminal configuration interface.

[0081] When the edge-level linkage drive value exceeds the first drive threshold, a high-level warning signal is output to the vehicle instrument panel via the vehicle communication link. The vehicle communication link, based on the CAN bus, completes the transmission of high-priority warning information, illuminating a red indicator light at the vehicle end and triggering a continuous buzzer warning. Simultaneously, the vehicle voice component outputs safety prompts. A complete warning message, including the high-level warning status, risk source analysis results, and emergency response guidelines, is pushed to the mobile terminal via the wireless communication link. The wireless communication link uses a cellular communication network and achieves high real-time push via MQTT and WebSocket protocols. On the cloud side, the edge side uploads battery status data, corresponding warning event numbers, and analysis results in real time to form a traceable high-risk event record. At the same time, after-sales personnel can remotely configure and adjust the sensor sampling frequency and analysis strategy on the edge side through the cloud platform.

[0082] In this implementation plan, based on the real-time changes of edge-level linkage driving values, a set of early warning and interaction mechanisms with hierarchical triggering, terminal response, and scenario linkage is constructed, so that battery operation risks can be perceived, alerted, and handled in a matched manner at different levels of severity. By comparing the linkage driving value with the two-level driving threshold, a tiered judgment of risk level is achieved. In the low-risk stage, the operational status is visualized and data is continuously recorded. As the risk worsens, audible and visual prompts and behavioral guidance are gradually introduced. In the high-risk stage, strong alerts and emergency information output are triggered, thus avoiding insufficient information or excessive interference caused by a single prompting method at different risk stages. Through the coordinated use of vehicle-mounted communication links, wireless communication links, and cloud communication mechanisms, warning information can be simultaneously disseminated between the vehicle, mobile device, and cloud, ensuring real-time risk perception during driving and supporting continuous status feedback for users in non-vehicle scenarios. Simultaneously, by opening different levels of parameter configuration and data upload strategies under different risk levels, sampling and analysis resources can be dynamically adjusted according to the risk level. This improves monitoring and analysis capabilities in the high-risk stage while controlling communication and computing load in the low-risk stage, thus forming an edge-level hierarchical linkage response system that balances real-time performance, reliability, and resource efficiency.

[0083] Specifically, the battery operating status is continuously assessed at the edge based on changes in behavioral load and risk evolution. The assessment results are then converted into actionable usage guidance and maintenance prompts before multi-terminal output and recording. The specific steps are as follows: At the edge, start-stop habit pulse load values, structural risk accumulation characterization values, and edge-level linkage drive values ​​are acquired. Simultaneously, electrochemical impedance spectroscopy (EIS) acquisition results and vehicle motion state data are read on the same time reference. The motion state data includes vehicle speed, path gradient distribution, and state changes during driving. By uniformly organizing and aligning these multiple data types in time, a multi-source state input set for service decision-making is formed, enabling subsequent service generation to simultaneously consider behavioral intensity, structural state, and actual operating conditions. Based on this, the current battery operating status is comprehensively analyzed based on the morphological changes of the EIS at different frequency bands. The internal structure and reaction state of the battery are evaluated from the aspects of conductivity response, charge transfer, and diffusion limitation, forming a battery health status assessment result. This battery health status assessment result serves as the basic reference for the generation of various subsequent services, reflecting the overall battery carrying capacity and sustainable operating level.

[0084] By combining the distribution characteristics of start-stop habit pulse load values ​​at different times, the real-time temperature status of the battery, and the battery health status assessment results, the matching degree between the user's existing charging behavior and the current battery load conditions is analyzed. When situations such as high start-stop load superposition or thermal state being unfavorable for charging are identified, targeted charging period adjustment suggestions are generated to guide charging behavior to maintain coordination with battery state. During vehicle operation, the electrical and thermal response states of the battery side during the energy recovery phase are analyzed based on the path slope distribution, driving speed, and real-time battery state changes. When continuous downhill and concentrated energy recovery scenarios are identified, corresponding driving and energy recovery guidance information is generated to avoid additional instantaneous load impacts on the battery during the energy recovery process.

[0085] Meanwhile, when the battery health status assessment results show a continuous downward trend across multiple operating phases, maintenance reminders are generated, and the corresponding status basis and analysis results are output as maintenance reference content to prompt users to perform battery testing and maintenance operations. Finally, the battery health assessment results, charging suggestions, driving guidance information, and maintenance reminders are synchronously output through the interaction link between the vehicle and mobile terminals, and the corresponding service decision records and status data are written to cloud storage to support subsequent status retrospective, trend analysis, and after-sales service arrangements, thereby forming a continuous service closed loop covering early warning, diagnosis, and maintenance.

[0086] In this implementation plan, the risk identification results are transformed into executable, interpretable, and continuously updatable service decision outputs at the edge, thereby achieving an effective extension from state awareness to user services and operation and maintenance support. By unifying and aggregating start-stop habitual pulse load values, structural risk accumulation characterization values, edge-level linkage driving values, and electrochemical impedance spectroscopy and vehicle motion state data, service decisions no longer rely on single indicators and instantaneous anomalies, but are based on multi-dimensional state collaborative analysis of behavior, structure, and operating conditions. By analyzing impedance spectrum morphology changes and generating battery health status assessment results, a stable and quantifiable reference benchmark is provided for charging recommendations, driving guidance, and maintenance reminders, avoiding frequent switching of service strategies due to short-term fluctuations. By combining start-stop behavior characteristics, temperature status, and health assessment results, charging period recommendations and energy recovery guidance can be dynamically matched to the current battery capacity and usage scenario, reducing the cumulative burden on the battery caused by improper use. At the same time, by triggering maintenance reminders when the health status continues to deteriorate and outputting relevant status data, users and after-sales personnel can clearly understand the reasons for maintenance recommendations. Finally, through collaborative output and storage on the vehicle, mobile, and cloud, a closed-loop mechanism is formed from risk identification to service guidance to status backtracking, improving overall vehicle safety, service targeting, and the traceability of subsequent operation and maintenance decisions.

[0087] The second aspect of this invention provides a vehicle-mounted battery multi-parameter edge warning device based on habit modeling, comprising: a multi-parameter synchronous acquisition and encapsulation module, used to uniformly perceive the battery operating status and vehicle operating conditions during vehicle operation, synchronously acquiring battery status data, battery feature data, and motion status data with a sensor array through a multi-parameter sensing interface, and performing preprocessing operations such as time alignment, noise reduction, anomaly removal, and missing data completion on the acquired multi-source data, thereby completing frame encapsulation based on a unified timestamp, thereby constructing a standardized vehicle status dataset with consistent structure that can be directly used for subsequent analysis and processing, providing a stable data input foundation for each analysis module on the edge side.

[0088] The pulse load characterization module is used to analyze the pulse load evolution characteristics of the battery during the start-stop process based on a standardized vehicle state dataset, focusing on the electrical fluctuations and thermal response changes caused by vehicle start-stop behavior. It transforms instantaneous start-stop disturbances into load characterization results with temporal continuity, and uses these load characterization results to depict the structural load state that the battery experiences in actual vehicle use scenarios. At the same time, it dynamically adjusts the processing intensity and data retention strategy of subsequent analysis links on the edge side according to the degree of load change, so that analysis resources are adaptively allocated according to the level of behavioral load.

[0089] The impedance cumulative change analysis module is used to continuously analyze the electrochemical impedance spectrum and its time-varying characteristics based on a standardized vehicle state dataset. It assesses the cumulative state of internal structural risks of the battery from multiple levels, including conductivity response, charge transfer, and diffusion limitation. The risk assessment results are used as structural state variables to reflect the potential degradation process inside the battery. At the same time, the depth and frequency of subsequent state analysis are corrected in real time based on the risk assessment results to obtain more sufficient structural evidence in the risk enhancement stage.

[0090] The edge-level early warning module is used to comprehensively evaluate the degree of linkage between battery behavior-side load and structural-side risk during vehicle operation by using pulse load characterization results and impedance cumulative change analysis results as joint inputs. Based on the linkage evaluation results, the module dynamically adjusts the level-level early warning strategy at the edge to match the early warning response intensity with the current risk level. At the same time, it initiates the corresponding service decision process when the trigger conditions are met.

[0091] The application service linkage decision module is used to continuously receive behavioral load change information and structural risk evolution status at the edge, periodically judge and update the battery operation status, and transform the judgment results into specific usage guidance suggestions, driving and charging prompts, and maintenance reminders. Synchronous output and recording are completed through multi-terminal interaction, thus forming a closed-loop processing flow from status perception, risk analysis to service response.

[0092] like Figure 4The diagram shows a cross-sectional view of the multi-layer PCB layout inside the terminal box provided in this application example. The overall design employs a partitioned layout and multi-layer shielding to support stable operation of multi-parameter acquisition, edge computing, and multi-link communication in an automotive environment. The terminal box shell forms a continuous, closed metal shielding cavity. Internally, grounding copper foil and grounding springs achieve shell-level electromagnetic continuity, suppressing external electromagnetic interference and limiting internal radiation leakage. Structurally, the interior is divided into multiple functional areas along the longitudinal direction. The upper area houses multiple communication module mounting positions, including 45mm×30mm and 25mm×30mm communication module slots for installing cellular communication, short-range communication, and positioning-related modules. Independent shielding cavities and grounding fences are set around the modules to reduce radio frequency crosstalk. The central area is the core control and storage area, employing a dual-controller structure and housing STM32H743VIT6 and STM32L476 devices. Compact routing and zoned power supply achieve coordinated operation of high-performance computing and low-power control. A continuous shielded grounding strip surrounds the main control area, with reserved spacing between controllers to meet heat dissipation and signal integrity requirements. The lower area is the underlying power and sensor interface area, housing EMI filtering components, power conditioning devices, and multi-channel sensor interface connections. The interface area is electromagnetically isolated from the main control area by a shielded isolation strip. Multiple EMI filtering components and shielded grounding structures are distributed in key routing areas of the entire circuit board to suppress power and high-speed signals, preventing interference propagation between different functional areas. Low-impedance return paths are formed between functional modules through shielded connectors and grounding springs, ensuring stable collaborative operation of the communication module, main control computing unit, and multi-parameter acquisition interface in complex automotive electromagnetic environments. The overall structure reflects an automotive edge computing hardware layout design with electromagnetic compatibility, functional isolation, and reliable heat dissipation as core objectives.

[0093] like Figure 5The diagram shows the circuit topology of the power management module provided in this application example. It converts the wide-range DC power supplied by the vehicle into multiple stable low-voltage power supplies, providing reliable power support for subsequent edge computing units, communication units, and sensing interfaces. The left side of the circuit is the vehicle power input terminal, supporting 12V to 24V vehicle DC input. A reverse connection protection circuit and transient suppression structure are arranged at the input terminal, using diodes and suppression devices to prevent damage to the internal circuitry caused by reverse connection, voltage surges, and load changes. After the power input, the circuit sequentially sets up a switching regulator unit and an inductor filter network to convert the high-voltage input into an intermediate regulated power rail. This part uses a filter structure composed of inductors, capacitors, and rectifiers to suppress switching noise, thereby reducing power ripple and electromagnetic interference. The intermediate regulated power rail is further distributed to multiple linear voltage regulator chips, each outputting a different level of stable DC voltage to meet the power supply requirements of different functional modules. The diagram on the right shows a multi-channel low-voltage output structure, including a 3.3V power output for the digital processing unit and communication module, a 5V power output for the interface and peripheral units, and a 12V regulated output for specific functional modules. Each regulated output is equipped with decoupling and filtering capacitors to further suppress high-frequency noise and improve load transient response. All voltage regulator chips employ a unified enable and ground reference structure to ensure stable and coordinated operation of multiple power supplies under complex automotive conditions. The overall circuit effectively isolates high-voltage fluctuations and interference from the automotive power supply through graded voltage regulation and multiple filtering, providing a stable and reliable power environment for the edge computing, signal acquisition, and communication modules within the automotive terminal, meeting long-term operation and electromagnetic compatibility requirements.

[0094] In this implementation plan, the role of the multi-parameter synchronous acquisition and encapsulation module is to establish a unified data perception and organization foundation at the vehicle end. By synchronously acquiring and preprocessing the battery electrical state, thermal state, internal structural response, and vehicle motion state, the originally scattered and asynchronous multi-source data is transformed into a time-aligned and structurally consistent standardized vehicle state dataset. This ensures that subsequent analysis can be based on the actual operation process and avoids distortion of behavioral analysis and structural analysis results due to asynchronous sampling, noise interference, and data loss.

[0095] The role of the pulse load characterization module is to transform the instantaneous electrical and thermal disturbances caused by vehicle start-stop behavior into a load evolution characterization with time continuity, so that the frequent start-stop and low-speed high-load usage habits can be amplified and explicitly expressed at the numerical level, thereby revealing the structural load state of the battery under actual commuting scenarios. The load characterization results are used as the basis for edge-side analysis link adjustment, so that analysis resources can be dynamically adjusted according to the intensity of behavioral load.

[0096] The role of the impedance accumulation change analysis module is to characterize the long-term evolution of internal risks in the battery from a structural level. By continuously analyzing the characteristics of electrochemical impedance spectroscopy changes over time, it identifies abnormal accumulation trends in conductive paths, charge transfer and diffusion processes, enabling internal structural degradation to be detected in advance before surface parameters show obvious abnormalities. Based on this, the depth and frequency of subsequent state analysis are dynamically adjusted, improving the foresight of structural risk identification.

[0097] The role of the edge-level early warning module is to comprehensively evaluate the linkage between behavioral load changes and structural risk accumulation, integrate the multi-dimensional analysis results into a linkage status that can be directly used for risk classification judgment, and dynamically adjust the edge-side early warning response strategy according to the degree of linkage, so that the early warning output neither over-amplifies short-term fluctuations nor ignores continuous risk accumulation, thereby achieving a reasonable match between risk perception and response rhythm.

[0098] The role of the application service linkage decision module is to transform the load analysis results, risk assessment results, and early warning levels generated on the edge side into specific service outputs for users and the operation and maintenance side. By generating executable information including charging suggestions, driving guidance, and maintenance reminders, the risk identification results can directly guide actual use and maintenance behavior. Through multi-terminal output and recording, a service closed loop is formed, which improves battery use safety, service targeting, and the traceability of subsequent operation and maintenance decisions.

[0099] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0100] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to the specific implementations described. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A multi-parameter edge warning method for vehicle batteries based on habit modeling, characterized in that: include: S1: Collect battery status data, battery characteristic data, and motion status data during vehicle operation; preprocess the battery status data, battery characteristic data, and motion status data to construct a standardized vehicle status dataset. S2, based on a standardized vehicle state dataset, performs load analysis on the battery structural load state from the battery pulse load evolution characteristics caused by vehicle start-stop behavior, and drives the dynamic adjustment of the edge-side analysis link based on the load analysis results; S3, based on a standardized vehicle state dataset, performs risk assessment on the cumulative state of structural risks inside the battery, and adjusts the depth of subsequent state analysis in real time based on the assessment results. S4 takes load analysis results and risk assessment results as input to conduct a linkage assessment of the linkage degree of vehicle battery operation risk, and drives the adjustment of edge-side hierarchical early warning strategy and service decision triggering based on the linkage assessment results; S5 continuously assesses battery operation status at the edge based on changes in behavioral load and risk evolution, and then converts the assessment results into executable usage guidance and maintenance prompts before completing multi-terminal output and recording.

2. The method for multi-parameter edge warning of vehicle battery based on habit modeling according to claim 1, characterized in that: The specific steps for collecting battery status data, battery feature data, and motion status data during vehicle operation, and preprocessing the battery status data, battery feature data, and motion status data to construct a standardized vehicle status dataset are as follows: The battery status data during vehicle operation is collected through a multi-parameter sensing interface module and sensor array deployed inside the battery. The battery status data includes: cell voltage value, cell temperature value, module temperature value, battery pack current and battery pack voltage at each sampling time. By using pressure sensors installed in the battery pack exhaust channel and inside the casing, as well as a dedicated EIS interface, battery characteristic data reflecting the internal structure and reaction state of the battery are collected. The battery characteristic data includes: internal pressure value of the battery pack, concentration of characteristic gases, and high-frequency complex impedance value, mid-frequency complex impedance value and low-frequency complex impedance value in the electrochemical impedance spectrum generated by the impedance analyzer. By integrating GPS and Beidou dual-mode positioning units with a six-axis inertial navigation component, kinematic state data during vehicle operation is obtained. The kinematic state data includes: the slope distribution of the road the vehicle is traveling on, the driving speed, and the change in the vehicle's steering angle. Record the number of sampling points contained in each sampling time and sliding time window, and divide a complete start, drive, brake and idle process into a start-stop segment, and record the start time and end time of each start-stop segment; Multi-level noise reduction and outlier removal strategies are employed for the collected battery status data, battery feature data, and motion status data: high-frequency noise from the sensor is filtered using a sliding window filtering algorithm; outliers are removed using the 3σ criterion; missing data is filled in using a data completion algorithm based on linear interpolation; the standardized battery status data, battery feature data, and motion status data are normalized by extreme value linear normalization based on a sliding time window, and then framed in chronological order. Each frame contains a timestamp, parameter type identifier, value, acquisition precision identifier, and acquisition channel number. Between frames, the data is written to a hierarchical buffer consisting of eMMC and microSD in ascending order of timestamp, thus constructing a standardized vehicle status dataset.

3. The method for multi-parameter edge warning of vehicle battery based on habit modeling according to claim 1, characterized in that: The specific steps for load analysis of battery structural load state based on standardized vehicle state dataset and battery pulse load evolution characteristics caused by vehicle start-stop behavior are as follows: The differences between the cell voltage value at the start of the start-stop segment and the cell voltage values ​​at each time point are calculated and summed to obtain the cumulative voltage pulse value. The temperature change density is obtained by taking the difference between the maximum and minimum cell temperatures in the start-stop segment as the numerator, and the difference between the start and stop times of the start-stop segment and the sum of the end and start times of the start-stop segment as the denominator. The difference between the internal pressure value of the battery pack at each moment in the start-stop segment and the internal pressure value of the battery pack at the beginning of the start-stop segment is calculated and summed to obtain the cumulative pressure hysteresis value. The difference between the characteristic gas concentration value at each moment in the start-stop segment and the characteristic gas concentration value at the start moment of the start-stop segment is calculated and summed to obtain the gas hysteresis cumulative value. The sum of the pressure hysteresis cumulative value and the gas hysteresis cumulative value is calculated, and then multiplied by the voltage pulse cumulative value and the temperature change density to obtain the multi-feature joint value; The start-stop habitual pulse load value is obtained by summing the combined values ​​of the multiple features of all start-stop segments and dividing by the number of sampling points.

4. The method for multi-parameter edge warning of vehicle battery based on habit modeling according to claim 1, characterized in that: The specific steps for dynamically adjusting the edge-side analysis link based on load analysis results are as follows: On the edge side, the start-stop habit pulse load values ​​obtained in the continuous commuting scenario are organized into a time series using a sliding time window method to form a start-stop habit pulse load evolution sequence, and the growth rate, fluctuation range and persistence characteristics of the start-stop habit pulse load evolution sequence on the time axis are recorded. The edge side compares the current start-stop habit pulse load value with the start-stop habit pulse load evolution sequence formed by the same vehicle within the historical time window, and calculates the relative position and offset trend of the start-stop habit pulse load value in the historical distribution. When the start-stop habitual pulse load value shows a monotonically increasing trend over N consecutive time windows, the edge-triggered adaptive enhancement measures for subsequent analysis processes include increasing the impedance spectrum sampling trigger frequency, extending the statistical window length for temperature gradient and pressure changes, and retaining key data segments with higher temporal resolution in the local cache. Meanwhile, the start / stop habit pulse load value and offset trend are written as behavioral inputs into the edge-side risk analysis context and stored in association with existing battery status data. Meanwhile, the edge side outputs the offset trend of start / stop habit pulse load values ​​to the vehicle and mobile terminals through a multi-protocol communication link.

5. The method for multi-parameter edge warning of vehicle battery based on habit modeling according to claim 1, characterized in that: The specific steps for risk assessment of the cumulative state of internal structural risks of the battery based on the standardized vehicle state dataset are as follows: Divide the peak-valley difference of cell voltage within the start-stop segment by the maximum cell temperature value within the start-stop segment to obtain the start-stop cell load value. Calculate the difference between the maximum cell temperature and the minimum cell temperature in the start-stop segment, and then divide it by the difference between the maximum module temperature and the minimum module temperature in the start-stop segment to obtain the thermal distribution imbalance characterization value. Calculate the ratio of the absolute value of the high-frequency complex impedance value to the absolute value of the mid-frequency complex impedance value in the impedance spectrum, and then take the absolute value of the ratio to obtain the impedance dissipation frequency band expansion value. Calculate the absolute value of the difference between the imaginary part of the low-frequency complex impedance value and the imaginary part of the mid-frequency complex impedance value, and then divide it by the value of the absolute value of the difference between the real part of the high-frequency complex impedance value and the real part of the mid-frequency complex impedance value plus one to obtain the diffusion-dominated shift characterization value. Multiply the start-stop cell load value, the thermal distribution imbalance characterization value, the impedance dissipation frequency band expansion value, and the diffusion-dominant offset characterization value to obtain the intermediate value of structural coupling expansion; The cumulative structural risk characterization value is obtained by adding one to the intermediate value of the structural coupling expansion and taking the natural logarithm.

6. The method for multi-parameter edge warning of vehicle battery based on habit modeling according to claim 1, characterized in that: The specific steps for real-time correction of the subsequent state analysis depth based on the evaluation results are as follows: After obtaining the cumulative structural risk characterization value, the cumulative structural risk characterization value is written into the edge-side operation status table as a structural state index, and a one-to-one mapping relationship is established with the electrochemical impedance spectroscopy, battery state data, and internal pressure value record of the battery pack corresponding to the current analysis window, forming a state index sequence with structural state as the main line; for the electrochemical impedance spectroscopy sampling results newly entering the analysis window, the edge side extracts the high-frequency complex impedance value, mid-frequency complex impedance value, and low-frequency complex impedance value respectively, and compares them segment by segment with the electrochemical impedance spectroscopy sampling results of the corresponding frequency band in the adjacent analysis window. When each frequency band shows the same direction of shift in the same window and the shift amplitude keeps changing continuously between adjacent windows, the current analysis window is marked as a significant structural change window, and the cell voltage change trajectory, internal pressure change record of the battery pack, and temperature distribution data corresponding to the significant structural change window are independently sealed and stored; When the cumulative structural risk characterization value maintains a monotonically increasing characteristic within M consecutive windows, the electrochemical impedance spectroscopy analysis path is switched, and the conventional full-band scanning results are split into high-frequency conductivity response segment, mid-frequency charge transfer segment and low-frequency diffusion-restricted segment for separate storage, and the index identifier corresponding to the cumulative structural risk characterization value is added to the storage record. In subsequent analysis, when changes in the electrochemical impedance spectroscopy morphology are detected, the corresponding historical structural evolution fragments are retrieved first through the structural risk accumulation characterization value index. The start-stop habitual pulse load values, cell voltage value change trajectories, and internal pressure value change records under the same index are retrieved. The state evolution sequence before and after the structural change is aligned and compared to determine whether the structural change belongs to a stage in the continuous structural risk accumulation process. The determination result is recorded as a structural risk evolution path and written into the edge-side event log.

7. The method for multi-parameter edge warning of vehicle battery based on habit modeling according to claim 1, characterized in that: The specific steps for assessing the degree of linkage between load analysis results and risk assessment results as inputs to evaluate the operational risks of vehicle batteries are as follows: Add one to the start / stop habitual pulse load value and take the natural logarithm to obtain the start / stop pulse load compression value; The cumulative structural risk value is obtained by adding one to the cumulative structural risk value and taking the natural logarithm. Divide the battery pack current by the battery pack voltage to obtain the instantaneous electrical response characterization value; The arctangent mapping is performed by multiplying the start-stop pulse load compression characterization value, the structural risk cumulative compression value, and the instantaneous electrical response characterization value as independent variables, and then multiplying by 2π to obtain the edge-level linkage driving value.

8. The method for multi-parameter edge warning of vehicle battery based on habit modeling according to claim 1, characterized in that: The specific steps for adjusting the edge-side hierarchical early warning strategy and triggering service decisions based on the linkage evaluation results are as follows: Real-time comparison of the current edge-level linkage driving value with the linkage driving threshold. The linkage driving threshold includes a first driving threshold and a second driving threshold, wherein the first driving threshold is greater than the second driving threshold. When the edge-level linkage drive value is less than or equal to the second drive threshold, the low-level warning status and current battery operating status data are sent to the vehicle instrument panel through the vehicle communication link. The corresponding low-risk warning indicator light is illuminated on the vehicle instrument panel and remains in silent warning mode. At the same time, the operation information containing the current battery status summary is pushed to the mobile terminal through the wireless communication link. Battery status records and habit trajectory analysis results are periodically uploaded to the cloud side to form daily operation reports. Users can configure and adjust the sensor sampling frequency and data upload cycle through the mobile terminal and cloud interface. When the edge-level linkage drive value is greater than the second drive threshold and less than or equal to the first drive threshold, a medium-level warning signal is output to the vehicle instrument through the vehicle communication link. An orange indicator light illuminates at the vehicle end and triggers an intermittent buzzer. At the same time, a warning message containing the warning level, risk source analysis information, and operation precautions is pushed to the mobile terminal through the wireless communication link. It also supports one-click sharing of warning information to the car manufacturer's after-sales and rescue platform through the message queue protocol. The warning log is recorded synchronously on the cloud side and a risk trend analysis task is started. At the same time, the interface for adjusting the sampling frequency and warning-related parameters by the mobile terminal and the cloud is opened. When the edge-level linkage drive value exceeds the first drive threshold, a high-level warning signal is output to the vehicle instrument through the vehicle communication link. A red indicator light illuminates at the vehicle end and triggers a continuous buzzer. At the same time, the vehicle voice component outputs safety prompt information. A complete warning message, including the high-level warning status, risk source analysis results, and emergency response guidelines, is pushed to the mobile terminal through the wireless communication link. Battery status data, warning event number, and analysis results are uploaded to the cloud in real time. At the same time, after-sales personnel can remotely configure the sampling frequency and analysis strategy through the cloud platform.

9. The multi-parameter edge warning method for vehicle batteries based on habit modeling according to claim 1, characterized in that: The specific steps for continuously judging the battery operation status at the edge based on changes in behavioral load and risk evolution, and then converting the judgment results into executable usage guidance and maintenance prompts to complete multi-terminal output and recording are as follows: The system acquires start-stop habit pulse load values, structural risk accumulation characterization values, and edge hierarchical linkage drive values ​​at the edge side, and simultaneously reads electrochemical impedance spectroscopy acquisition results and vehicle motion state data to form a multi-source state input set for service decision-making. Based on the morphological changes of electrochemical impedance spectroscopy, the current operating status of the battery is comprehensively analyzed to form a battery health status assessment result, which is then used as the basic reference for subsequent service generation. By combining the distribution characteristics of start-stop habit pulse load values ​​in different time periods, battery temperature status, and battery health status assessment results, the degree of matching between the user's existing charging behavior and battery load conditions is analyzed, and charging time adjustment suggestions are generated. During vehicle operation, the electrical and thermal response states of the battery side during the energy recovery phase are analyzed based on the path slope distribution, driving speed, and real-time battery status changes. When a continuous downhill scenario is identified, corresponding driving and energy recovery guidance information is generated. When the battery health status assessment results show a continuous downward trend, a maintenance reminder message is generated, and the corresponding status basis and analysis results are output as maintenance reference content to prompt the user to perform battery testing and maintenance operations. Battery health assessment results, charging recommendations, driving guidance information, and maintenance reminders are output synchronously through the interaction link between the vehicle and the mobile terminal, and the corresponding service decision records and status data are written to cloud storage.

10. A vehicle battery multi-parameter edge warning device based on habit modeling, employing the vehicle battery multi-parameter edge warning method based on habit modeling as described in any one of claims 1-9, characterized in that: include: The multi-parameter synchronous acquisition and encapsulation module is used to collect battery status data, battery feature data, and motion status data during vehicle operation, and to preprocess the battery status data, battery feature data, and motion status data to construct a standardized vehicle status dataset. The pulse load characterization module is used to perform load analysis on the battery structural load state based on the battery pulse load evolution characteristics caused by vehicle start-stop behavior, based on a standardized vehicle state dataset, and drive the dynamic adjustment of the edge-side analysis link based on the load analysis results. The impedance cumulative change analysis module is used to assess the cumulative state of internal structural risks of the battery based on a standardized vehicle state dataset, and to correct the depth of subsequent state analysis in real time based on the assessment results. The edge-level early warning module is used to assess the degree of linkage of vehicle battery operation risks based on load analysis results and risk assessment results, and to drive the adjustment of edge-side level early warning strategies and service decision triggering based on the linkage assessment results. The application service linkage decision module is used to continuously judge the battery operation status at the edge based on changes in behavioral load and risk evolution, and then convert the judgment results into executable usage guidance and maintenance prompts before completing multi-terminal output and recording.

Images