Battery self-discharge characteristic analysis and rapid evaluation method

Abstract

The invention relates to a battery self-discharge characteristic analysis and rapid evaluation method, and aims to improve the precision and efficiency of battery self-discharge performance detection. According to the method, in a set environment stable interval, an instantaneous voltage change sequence and a current perturbation response sequence of a target battery are acquired through micro-load disturbance, and a battery internal resistance dynamic evolution tensor representing dynamic response characteristics in the battery is constructed; extracting local features by applying a multi-scale time window segmentation strategy, forming a feature vector group containing a voltage nonlinear rebound factor and a delay response index, inputting the feature vector group into a trained Bayesian weighted adaptive mapping network, and outputting a battery self-discharge rate estimated value combining a manufacturing batch residual error and temperature offset compensation; performing trend fitting and consistency checking on the estimated value to obtain a battery self-discharge grade evaluation result, and generating a structured detection report containing initial offset, rebound amplitude and estimation confidence; the method can be widely applied to battery production, quality control and performance screening scenes.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of batteries, in particular to a battery self-discharge characteristic analysis and rapid evaluation method. BACKGROUND

[0002] In the prior art, the evaluation of battery self-discharge characteristics mainly relies on long-time static observation method, that is, the voltage decay of the battery is monitored for several days to several weeks under constant temperature conditions, so as to estimate the self-discharge rate. In addition, some methods indirectly infer the self-discharge trend through static impedance measurement or open-circuit voltage comparison. These methods are mainly used for production line sampling or laboratory verification link, and have certain theoretical basis and engineering feasibility.

[0003] However, the above-mentioned prior art generally has problems of long detection period, poor real-time performance, strong dependence on environmental conditions, etc., and is difficult to adapt to the needs of rapid screening and refined evaluation in modern battery mass production. At the same time, some indirect inference methods lack the modeling ability of dynamic response behavior, which limits the accuracy and stability of the evaluation results, especially the performance fluctuations caused by manufacturing process differences or temperature drift.

[0004] Therefore, it is urgent to propose a new method for analyzing and evaluating the self-discharge characteristics of batteries with high precision in a short time. SUMMARY

[0005] The present application provides a battery self-discharge characteristic analysis and rapid evaluation method to improve the precision and efficiency of battery self-discharge performance detection.

[0006] The present application provides a battery self-discharge characteristic analysis and rapid evaluation method, comprising: In a set environmental stability interval, a plurality of groups of instantaneous voltage change sequences and current micro-perturbation response sequences of a target battery under micro-load disturbance in short time segments are obtained; Based on the instantaneous voltage change sequence and the current micro-perturbation response sequence, a battery internal resistance dynamic evolution tensor for representing the internal dynamic response characteristics of the battery is constructed; The battery internal resistance dynamic evolution tensor is subjected to local feature extraction by applying a multi-scale time window segmentation strategy, and a feature vector group containing a voltage nonlinear rebound factor and a delay response index is obtained; The feature vector group is input into a trained Bayesian weighted adaptive mapping network, and a battery self-discharge rate estimate value considering manufacturing batch residual modeling and temperature offset compensation mechanism is obtained; The battery self-discharge rate estimate values of at least three consecutive periods are subjected to trend curve fitting, and the consistency is tested, and a battery self-discharge grade evaluation result is obtained; Based on the battery self-discharge level evaluation result and the corresponding battery self-discharge rate estimation value, a structured detection report containing a starting offset, a rebound amplitude, and an estimation confidence is generated.

[0007] Further, the micro-load disturbance refers to applying a short-time pulse current signal with an amplitude less than 1% of the rated capacity to the target battery during the test, and each disturbance lasts less than 2 seconds to avoid significantly changing the state of charge of the battery.

[0008] Further, the voltage nonlinear rebound factor represents the degree of deviation of the voltage recovery process after the disturbance from the linear trend, which is obtained by fitting the voltage sequence in the disturbance recovery segment and calculating the average value of the residual squares; the delay response index is used to reflect the response time delay of the voltage after the current disturbance is applied, and the value is the time interval between the current transition time and the time when the voltage maximum change rate occurs, and is normalized as the proportion of the length of the disturbance time window.

[0009] Further, the Bayesian weighted adaptive mapping network includes a feature encoding unit, a manufacturing batch and temperature condition joint modeling unit, and an output fusion and confidence reasoning unit. The feature encoding unit is configured to receive a normalized feature vector group, the feature vector group including a plurality of voltage nonlinear rebound factors and delay response indexes, and perform feature extraction and compression on the feature vector group through a multi-layer feedforward network to output a fixed-length encoding vector representing the disturbance response behavior characteristics. The manufacturing batch and temperature condition joint modeling unit is configured to receive manufacturing batch and environmental temperature information of the target battery, and concatenate the fixed-length encoding vector to generate a joint input vector; includes a plurality of sub-mapping paths, each of which is optimized based on training data of a specific manufacturing batch and temperature interval, and in the inference process, a Bayesian posterior weight is assigned according to the similarity between the current input condition and the center condition vector of each sub-path. The output fusion and confidence reasoning unit is configured to receive candidate battery self-discharge rate estimation values output by all sub-mapping paths, and combine the Bayesian posterior weight to weight and fuse each candidate value to generate a final battery self-discharge rate estimation value; and calculate a confidence score based on the dispersion degree between the candidate values to represent the reliability of the final battery self-discharge rate estimation value.

[0010] Further, based on the instantaneous voltage change sequence and the current micro-disturbance response sequence, a battery internal resistance dynamic evolution tensor representing the dynamic response characteristics of the battery is constructed, including: Within the set environmental stable interval, the instantaneous voltage variation sequence and the current perturbation response sequence of each group of short time segments are respectively divided into a pre-perturbation static segment, a perturbation excitation segment and a perturbation end recovery segment, and the segment division is automatically determined according to the initial slope and amplitude threshold of the current variation to determine the boundary time; In each perturbation excitation segment, the time relationship between the voltage drop starting point, the voltage minimum point and the current transition point corresponding to the current perturbation is extracted and marked as a voltage response window index field as a time positioning reference for subsequent tensor construction; In each perturbation end recovery segment, the recovery speed vector between consecutive voltage sampling points in the segment is calculated, and the recovery initial offset value, the maximum rebound speed and the rebound amplitude are combined to construct a three-dimensional battery internal resistance dynamic evolution tensor including time steps, three physical characteristics and segment numbers; The maximum rebound speed and the delay take-off time difference of each segment in the battery internal resistance dynamic evolution tensor are taken as a local dynamic feature field to participate in the subsequent operation of local feature extraction by using a multi-scale time window segmentation strategy.

[0011] The beneficial effects of the technical scheme provided by the application include: (1) By obtaining the instantaneous voltage and current response sequence under the micro-load perturbation, and combining the Bayesian weighted mapping network for modeling and estimation, the battery self-discharge rate can be quickly obtained without long static time, which greatly improves the detection efficiency and is suitable for rapid screening in the production link. (2) The battery internal resistance dynamic evolution tensor, the multi-scale time window segmentation strategy and the feature vector group construction can effectively extract the key parameters reflecting the micro changes in the battery, enhance the modeling ability of complex dynamic behavior, and improve the accuracy and stability of the self-discharge evaluation result. (3) By considering the manufacturing batch residual modeling and temperature offset compensation mechanism, the system can effectively eliminate the evaluation deviation caused by batch difference and environmental factors, and ensure the universality and reliability of the algorithm under the condition of multiple scenes and multiple specifications of batteries. BRIEF DESCRIPTION OF DRAWINGS

[0012] Figure 1 is a flowchart of a battery self-discharge characteristic analysis and rapid evaluation method provided by the first embodiment of the application. DETAILED DESCRIPTION

[0013] In the following description, many specific details are set forth in order to provide a thorough understanding of the application. However, the application can be practiced in many different ways beyond the specific details disclosed herein, and the skilled person can make similar extensions without departing from the scope of the application, so the application is not limited to the specific implementation disclosed below.

[0014] The first embodiment of the application provides a battery self-discharge characteristic analysis and rapid evaluation method. Please refer toFigure 1 , which is a schematic diagram of the first embodiment of the present application. The following will be described in combination with Figure 1 A battery self-discharge characteristic analysis and rapid evaluation method is provided for the first embodiment of the present application.

[0015] Step S101: In a set environment stable interval, a plurality of groups of short time segment instantaneous voltage change sequences and current micro-perturbation response sequences of the target battery under micro-load disturbance are obtained.

[0016] In step S101, the system first needs to construct an environment stable interval, which refers to a physical space in which the external conditions such as temperature, humidity, electromagnetic interference and airflow are basically constant and do not produce significant disturbance to the battery behavior. Preferably, the environment can be realized by a thermostat or a laboratory controlled test cabin, the temperature is maintained at 25±1℃, the relative humidity is controlled between 40% and 60%, there is no obvious airflow vibration and electromagnetic field interference outside, and the stability and repeatability of the data during the battery test are ensured.

[0017] Under the environment, the target battery is connected to a set of electronic load modules with milliamperes level resolution and programmable control capability, the electronic load should support constant current mode and can stably apply current disturbance within milliseconds. Preferably, the electronic load is controlled by a microcontroller or a host computer (such as STM32 or LabVIEW platform), a small amplitude current disturbance is set within a specified time window, for example, 0.5 mA, 1.0 mA or 2.0 mA. In order to avoid activating the battery macroscopic reaction or affecting the state of charge, the disturbance current should be much smaller than the working current corresponding to the rated capacity of the battery, and the disturbance duration is recommended to be set between 0.5 seconds and 2 seconds. A no-load segment of not less than 0.2 seconds should be set before and after the disturbance, respectively, for observing the natural voltage change and the recovery characteristics after the disturbance, thereby constituting a complete disturbance cycle.

[0018] During the application of current disturbance, the voltage change and the actual current flowing through the battery are measured synchronously by a set of high-resolution, high-sampling-rate data acquisition modules (such as 16-bit ADC modules, differential amplifiers and high-speed sampling systems). The acquisition system should support a sampling rate of not less than 1 kHz to ensure the capture accuracy of the transient process. The obtained data include voltage and current signal sequences marked with a unified time stamp, which are respectively referred to as instantaneous voltage change sequence and current micro-perturbation response sequence, both of which should be strictly aligned on the time axis and have equal number of data points.

[0019] In order to improve the signal-to-noise ratio and ensure the data quality, it is recommended to use wavelet transform denoising, Savitzky-Golay filtering or mean smoothing signal processing algorithm to preliminarily clean the original voltage and current data after acquisition. The key mutation points and gradual change characteristics of the signal should be preserved during the processing to avoid feature loss caused by excessive smoothing.

[0020] In addition, to achieve high reliability data acquisition, it is recommended that each group of disturbance tests should collect no less than 5 independent short time segments (which can be completed across several minute intervals), each segment should include a complete cycle of "rest-disturb-recovery". The system needs to evaluate the quality of each cycle of acquisition, if the voltage mutation is discontinuous, sampling jitter or disturbance response is abnormal, it should be marked as invalid and automatically excluded to prevent false data from entering the subsequent analysis process.

[0021] Finally, the output result of step S101 is a set of short time segment data with unified format, time synchronization and high signal-to-noise ratio, including instantaneous voltage change sequence and current micro-disturbance response sequence, as the basic input data for the subsequent step of constructing battery internal resistance dynamic evolution tensor. This step realizes high-precision capture of the micro-electrochemical response behavior of the battery through the construction of non-destructive, low-disturbance and standardized data acquisition mechanism, taking into account the test efficiency and accuracy, and ensuring the repeatability and industrial applicability of the entire evaluation process.

[0022] Further, the micro-load disturbance refers to applying a short-time pulse current signal with an amplitude less than 1% of the rated capacity to the target battery during the test, and the duration of each disturbance is less than 2 seconds, so as to avoid significantly changing the state of charge of the battery.

[0023] In the present application, the "micro-load disturbance" is used to stimulate the short-time voltage response characteristics of the battery, so as to extract the dynamic characteristic parameters of the battery internal resistance without destroying its state of charge. In order to ensure that such disturbance does not significantly change the state of charge of the target battery or trigger irreversible electrochemical changes, the amplitude and duration of the disturbance process must meet strict boundary conditions at the same time according to the present embodiment.

[0024] Specifically, in actual tests, the system controls the current source or electronic load device to apply a short-time current pulse to the battery, and the current amplitude of each pulse should be less than 1% of the rated discharge capacity of the battery. For example, for a lithium ion battery with a capacity of 2000 mAh, the upper limit of the allowed micro-disturbance current is 20 mA. The current can be implemented by a constant-current electronic load, a precision programmable DC source or by a digital analog control circuit, ensuring that the control accuracy is within ±0.1 mA.

[0025] In addition, in order to further reduce the influence of the disturbance on the polarization state and electrochemical stability of the battery, the duration of each pulse should not exceed 2 seconds. The preferred range is 0.5 seconds to 1.5 seconds, so as to ensure that the disturbance can cause measurable voltage changes while avoiding the activation of internal charge migration or interface reactions in the battery. The time control can be set by a microcontroller trigger and stop logic, or automatically executed by a test program to ensure high repeatability of each disturbance.

[0026] The entire micro-perturbation process needs to be performed in the environmental stable interval, that is, the environmental conditions such as temperature, humidity, electromagnetic interference remain constant during the test process, so as to exclude the interference of external factors on the perturbation response data. At least 200 milliseconds of static window should be left before and after the perturbation, for collecting the baseline voltage before the perturbation and the voltage recovery behavior after the end of the perturbation.

[0027] Through the above amplitude and time double constraints, the existing precision electronic load or test system can be used to implement non-destructive and repeatable perturbation test operation on any type of battery, so as to collect sufficient response information for subsequent analysis without affecting the performance of the battery. Such micro-perturbation control mechanism can be widely applied to lithium ion, nickel hydrogen, lead acid and other types of electrochemical energy storage devices.

[0028] Step S102: based on the instantaneous voltage change sequence and the current micro-perturbation response sequence, a battery internal resistance dynamic evolution tensor for representing the battery internal dynamic response characteristics is constructed.

[0029] In step S102, the system needs to construct a tensor structure for representing the internal electrochemical dynamic change process of the battery based on the instantaneous voltage change sequence and the current micro-perturbation response sequence collected in step S101. The construction process of the tensor not only includes the integration of the basic voltage and current corresponding relationship, but also needs to be expanded through the time dimension, the perturbation period dimension and the signal feature dimension, so as to form a time sequence tensor capable of carrying the multi-dimensional characteristics of the micro-perturbation response behavior.

[0030] Specifically, each group of voltage and current sequences contains multiple sampling points. The system first needs to align the voltage and current signals point by point to ensure that the two signals have physical response correspondence at the same time point. On this basis, the system divides a complete perturbation period into several time periods, such as the pre-perturbation static segment, the perturbation excitation segment, and the post-perturbation recovery segment, and extracts local voltage slope, current response amplitude, recovery speed and other key physical parameters in each time period. For each perturbation segment, the system can encode it into a three-dimensional array, where the first dimension is the time step, the second dimension is the signal type (voltage, current, slope, rebound rate, etc.), and the third dimension is the perturbation sequence number or test batch information. By splicing multiple perturbation segments in a unified tensor structure, the system can establish a multi-dimensional data body containing multiple samples, multiple physical variables and their time evolution relationship.

[0031] In the actual process of constructing the dynamic evolution tensor of battery internal resistance, it is necessary to first segment the complete disturbance period in each perturbation test segment to reflect the dynamic response process of the battery to the small disturbance. Specifically, each disturbance period can be divided into three periods according to the application time of the current disturbance as the dividing point: a pre-disturbance static segment, a disturbance excitation segment, and a disturbance end recovery segment. The pre-disturbance static segment refers to the stable state interval before the current disturbance is applied, usually lasting for 200 to 500 milliseconds, for capturing the baseline level of the voltage; the disturbance excitation segment refers to the high response zone after the formal application of the micro-load current, with a duration consistent with the disturbance time, usually for 500 milliseconds to 2 seconds, at which time the voltage appears a significant drop or change, which is an important section for observing the behavior of the equivalent series resistance (ESR) of the battery; the disturbance end recovery segment is the stage of natural rebound of the voltage after the current is disconnected, lasting for 500 milliseconds to 1 second, for depicting the polarization recovery and interface charge migration effect.

[0032] After the division is completed, the system will respectively extract the features of the voltage and current data in each period. The pre-disturbance static segment is used to calculate the voltage mean as the reference voltage baseline; the disturbance excitation segment can calculate the instantaneous voltage drop amplitude, instantaneous current amplitude, and voltage slope (i.e. ΔU / Δt) as the indicators representing the internal resistance and polarization speed; the disturbance end recovery segment can analyze the maximum rebound speed, rebound amplitude, and fitting index of the voltage recovery curve in the voltage recovery process, which reflect the hysteresis and recovery capacity of the internal electrochemical system of the battery. All extracted features will be embedded as signal dimensions into the channels of the tensor structure, providing high-dimensional, time-located microscopic behavior data for subsequent models.

[0033] The maximum rebound speed refers to the maximum rate of voltage change per unit time in the voltage recovery process after the disturbance current is terminated. The calculation method is as follows: during the period from the end of the disturbance to the voltage tending to be stable, let the sampling period be , for each two adjacent voltage sampling points and , calculate the first-order difference , after traversing all the recovery segment sampling points, take the maximum value as the maximum rebound speed, usually in units of volts / second ( ). This value represents the instantaneous recovery capacity of the battery at the initial stage of polarization release.

[0034] The rebound amplitude refers to the total voltage uplift between the lowest voltage at the end of the disturbance and the termination point of the recovery segment. Let the lowest voltage at the end of the disturbance be , and the voltage at the end of the recovery be , then the rebound amplitude is defined as: ;

[0035] The unit is volt (V). This value reflects the battery's total ability to recover from a polarized state to a steady state.

[0036] The fitting exponent of the voltage recovery curve is used to characterize the dynamic process of voltage recovering to a steady state over time. This fitting typically takes the form of an exponential recovery function: ;

[0037] in To recover the voltage at any point in time within the segment, This represents the final stable voltage (which can be approximated as the voltage value at the end of the recovery segment). To restore the initial offset, indicating the time when the disturbance ends. At that time, the voltage relative to the final stable voltage The initial deviation. That is to say, This is the voltage difference between the voltage at the point where it begins to recover from the lowest point after the disturbance and the final recovered voltage. This value reflects the degree of polarization or transient voltage drop of the battery at the instant the disturbance ends. The larger the value, the greater the recovery required after the disturbance, and the more significant the internal impedance or polarization response. The unit is volts, consistent with the voltage signal itself.

[0038] The fitting exponent (also called the recovery rate constant) of the voltage recovery curve is expressed in units of 1000 kJ / m². By fitting the voltage recovery segment sampling points to the above functional form using the nonlinear least squares method, the fitting exponent can be obtained. The larger this value, the faster the voltage recovers and the smaller the hysteresis.

[0039] During tensor construction, normalization is preferably used to make all signal dimensions dimensionless to eliminate the offset effect caused by differences in sampling amplitude. For example, voltage values ​​can be normalized to the percentage change relative to the initial perturbation value, and current signals can be normalized to the relative response under the perturbation amplitude unit. Furthermore, to enhance the expressive power of the tensor in subsequent modeling tasks, the system can also introduce a set of derived feature dimensions, such as parameters like the location of local extrema, the time interval between peaks and valleys, and the mutation rate of the derivative at the reversal point, and embed them into the tensor as extended channels.

[0040] The constructed battery internal resistance dynamic evolution tensor not only records the instantaneous response of the battery to external stimuli under each perturbation, but also implicitly reflects the dynamic trajectory of complex mechanisms such as its internal polarization characteristics, charge migration rate, and interface impedance changes. This tensor will serve as the main input data structure for subsequent steps to further extract representative features and build a prediction model.

[0041] In order to improve the construction efficiency and stability, the system should encapsulate the tensor generation process as an automatically running program module, and ensure that all fragment processing processes adopt unified time alignment strategy, signal processing algorithm and feature coding format, so as to facilitate the consistent training and evaluation of subsequent models. The core of this step is to use the structured and multi-dimensional tensor expression method to convert the battery short-time disturbance response data from time series to information carrier with spatial pattern and statistical characteristics, thereby providing a solid data foundation for accurately describing the battery self-discharge behavior.

[0042] Further, the battery internal resistance dynamic evolution tensor for representing the internal dynamic response characteristics of the battery is constructed based on the instantaneous voltage change sequence and the current perturbation response sequence, including: In the set environment stable interval, the instantaneous voltage change sequence and the current perturbation response sequence of each group of short time segments are divided into pre-perturbation static segment, perturbation excitation segment and perturbation end recovery segment, and the segment division is determined according to the initial slope and amplitude threshold of current change to automatically determine the boundary time; In each perturbation excitation segment, the time relationship between the voltage drop starting point, the voltage minimum point and the current transition point corresponding to the current perturbation is extracted and marked as a voltage response window index field, which is used as a time positioning reference for subsequent tensor construction; In each perturbation end recovery segment, the recovery speed vector between the continuous voltage sampling points in the segment is calculated, and the recovery initial offset value, the maximum rebound speed and the rebound amplitude are combined to construct a three-dimensional battery internal resistance dynamic evolution tensor including time steps, three physical characteristics and segment numbers; The maximum rebound speed and the delay take-off time difference of each segment in the battery internal resistance dynamic evolution tensor are used as local dynamic feature fields to participate in the operation of local feature extraction by using a multi-scale time window segmentation strategy.

[0043] In the battery self-discharge characteristic analysis and rapid evaluation method of the application, the process of constructing the battery internal resistance dynamic evolution tensor for representing the internal dynamic response characteristics of the battery is the core step of extracting key disturbance behaviors from the instantaneous voltage change sequence and the current perturbation response sequence. In order to accurately model the detailed response mechanism of the battery under the condition of micro-perturbation, the system takes multiple short time segments as analysis units, and performs operations such as segment identification, behavior index extraction, structured organization and feature field injection in the set environment stable interval, and finally forms a three-dimensional tensor representation with time structure, behavior characteristics and segment identification ability.

[0044] Firstly, the system divides each group of short time segments into the corresponding transient voltage variation sequence and current perturbation response sequence. The division process is completed automatically by the program and does not rely on manual annotation. The division is based on the initial slope and amplitude threshold of the current signal variation. When the current waveform rapidly transitions from the steady state to the perturbation current level, the system detects that the rising or falling slope exceeds the set threshold, which is determined as the starting point of the perturbation excitation segment. When the current waveform falls back and stabilizes near the baseline, and the variation amplitude is lower than the set threshold, the system marks this position as the end of the perturbation. After the division, each segment is divided into a pre-perturbation static segment, a perturbation excitation segment, and a perturbation end recovery segment, which correspond to the static state before the perturbation, the response stage during the current loading, and the process of voltage gradually recovering after the perturbation is removed, respectively.

[0045] In each perturbation excitation segment that has been divided, the system further locates the starting point of voltage drop, the voltage minimum point, and the peak point of current variation rate, measures the relative time interval between them, and records the time relationship as the voltage reaction window index field. This field can be regarded as a characteristic time marker of the perturbation response, which is used to assist the relative positioning of the time step sequence in the tensor and the window alignment control in the subsequent process, to ensure that the voltage dynamic behavior has comparability in the position of the tensor during the tensor construction process.

[0046] Subsequently, the system focuses on each perturbation end recovery segment, i.e., the time domain interval in which the voltage climbs back to the stable level after the perturbation is removed. In this segment, the system extracts the difference between all consecutive voltage sampling points divided by the sampling interval to form a recovery speed vector, which is used to reflect the speed change trajectory of the voltage rise in this segment. At the same time, the system further calculates three key physical characteristics: one is the recovery initial offset value, i.e., the sinking amplitude of the voltage at the perturbation end relative to the steady-state voltage before the perturbation; the second is the maximum rebound speed, i.e., the maximum upward rate in the recovery speed vector; the third is the rebound amplitude, i.e., the amplitude between the voltage rising from the minimum point to the final stable value. These three physical indicators and the recovery speed vector together form a three-dimensional physical feature dimension.

[0047] After completing the above steps, the system organizes and constructs the tensor in a unified format. The first dimension of the tensor is the continuous time step, reflecting the high-frequency response trajectory inside the perturbation segment; the second dimension is the three-dimensional physical feature, including the recovery speed vector, the recovery initial offset value, the rebound amplitude, and the maximum rebound speed; the third dimension is the segment number dimension, used to identify which perturbation segment it belongs to, ensuring that different segment data can be processed independently or classified in batches in subsequent analysis. The tensor adopts a standard structured memory layout with backward propagable index information, which can support the layer-by-layer input of the neural network model.

[0048] In addition, in order to enable the subsequent multi-scale time window segmentation strategy to fully utilize the key dynamic mode of the disturbance recovery feature, the system extracts the maximum rebound speed and the delay take-off time difference of the recovery segment corresponding to each segment in the aforementioned tensor, and adds them to the local dynamic feature set in the tensor as an independent field. The delay take-off time difference refers to the time delay from the end of the current disturbance to the significant rebound of the voltage, reflecting the time lag of the battery polarization or diffusion mechanism response. These local dynamic feature fields not only enhance the tensor's own representation ability of the disturbance response dynamics, but also provide a structured basis for subsequent time window selection, feature extraction, and classification aggregation operations.

[0049] Through the above processing, the application can finely model the multi-segment response behavior of the target battery under the micro-perturbation condition without long-period placement or external large disturbance, and construct a three-dimensional battery internal resistance dynamic evolution tensor with clear physical meaning, stable time alignment, and various feature granularities, as the core input structure of the subsequent analysis process, ensuring that the self-discharge characteristic evaluation has responsiveness, stability, and engineering executability.

[0050] Step S103: applying a multi-scale time window segmentation strategy to the battery internal resistance dynamic evolution tensor to extract local features and obtain a feature vector group containing voltage nonlinear rebound factors and delay response indexes.

[0051] In step S103, the system takes the battery internal resistance dynamic evolution tensor as input, structures the disturbance response data contained therein, and extracts voltage nonlinear rebound behavior and voltage response delay features therefrom. In order to accurately capture the local response characteristics of the battery under different time scales, the system adopts a multi-scale time window segmentation strategy. "Multi-scale" means that for the same disturbance data, multiple time window lengths are set, such as 50 milliseconds, 100 milliseconds, 200 milliseconds, and 500 milliseconds, each length representing a different time resolution. The system slides the time window to extract segments from the data at each scale, usually in a partially overlapping manner, to ensure that the overall disturbance process is completely covered.

[0052] For each extracted time window segment, the system performs two types of feature analysis. The first one is the extraction of voltage nonlinear rebound factor, aiming to identify whether the voltage presents regular and uniform change, or sudden rebound, jitter or nonlinear distortion in this time period. The implementation method is that the system constructs a linear fitting curve within the current window, which simulates the trend of voltage change over time. Then the actual sampled voltage values are compared with the expected values of the fitting curve point by point, and the sum of squares of all deviations is calculated, and then divided by the total number of sampling points in the window, thus obtaining a value representing the "average deviation degree". If this value is close to zero, it means that the voltage change in this segment is close to linear; if the value is larger, it means that the voltage response contains obvious nonlinear components, such as rebound peak, hysteresis turning point or transient fluctuation, etc. This value is defined as the "voltage nonlinear rebound factor" in the system, with the unit of square volt, which is used to quantify the local nonlinear strength.

[0053] The second one is the response delay index of voltage to current disturbance, aiming to evaluate whether the battery can quickly produce voltage response after the application of micro-perturbation. The system first marks the position of the start of the disturbance in the current sequence, which is usually the time point when the current suddenly changes from zero to the set disturbance value. Then in the subsequent voltage data, the time point with the fastest voltage change rate is found, which is usually the starting inflection point of the voltage sharp decline or rebound. This point can be found by analyzing the speed of voltage change, i.e. the voltage difference between consecutive sampling points divided by the time interval, to find the time point where the maximum value is located. Then the system calculates the time interval between this time point and the start point of the current disturbance, obtaining a response delay value in seconds. In order to eliminate the influence of different time window scales on the absolute size of the value, the system divides this delay value by the total length of the current time window, obtaining a proportional value, which is defined as the "delay response index". For example, in a window lasting 200 milliseconds, if the voltage starts to respond 40 milliseconds later than the current disturbance, the index is 0.2.

[0054] The above two features, voltage nonlinear rebound factor and delay response index, are combined as a local feature vector of a time window. The system will repeat this process in multiple scales and multiple sliding time windows to obtain multiple local feature vectors. All feature vectors are combined in time and scale order to form a complete feature vector group, representing the time and scale domain behavior distribution of the battery disturbance response process. These vectors will be directly input into the subsequent self-discharge rate estimation model for identifying the self-discharge characteristics of different batteries.

[0055] The whole feature extraction process is automatically executed by the pre-set analysis program, with unified data interface and standardized sliding window strategy, having high repeatability and good batch processing capability.

[0056] Further, the voltage nonlinear rebound factor represents the deviation degree of the voltage recovery process after the disturbance from the linear trend, which is obtained by fitting the voltage sequence in the disturbance recovery section and calculating the average value of the residual squares; the delayed response index is used to reflect the time delay of the voltage response after the current disturbance is applied, and the value is the time interval between the current transition time and the time when the voltage maximum change rate occurs, and is normalized as the proportion of the length of the disturbance time window.

[0057] In the battery self-discharge characteristic analysis and rapid evaluation method described in the application, in order to extract the key characteristic parameters that can characterize the internal dynamic behavior of the battery from the disturbance response process, the system introduces two core indicators in the multi-scale time window, which are the voltage nonlinear rebound factor and the delayed response index. These two indicators are used to capture the nonlinear behavior in the voltage recovery process after the disturbance ends and the time lag of the voltage response to the current disturbance, which can effectively enhance the discrimination accuracy of the subsequent self-discharge rate estimation.

[0058] In the extraction process of the voltage nonlinear rebound factor, the system first locates the corresponding current disturbance end time in each time window, that is, the moment when the micro load is terminated. Then the system intercepts a recovery period voltage sequence after this time, usually with a length of several hundred milliseconds to one second. In this period, the system uses a first-order linear trend as a reference baseline, that is, the voltage signal in the entire recovery section is fitted as an optimal straight line. The fitting process can be realized by least squares method, which ensures that the linear trend line is closest to the actual voltage change trend. After the fitting is completed, the system squares the deviation between each voltage sampling point and the fitted value, and calculates the average value or total amount, and the average value of the deviation is the voltage nonlinear rebound factor. The unit of this factor is square volt, and the larger the value is, the more obvious the nonlinear fluctuation in the voltage recovery process is, which may be related to the interface capacitance delay, electrolyte migration block or local polarization reconstruction.

[0059] The extraction of the delay response index focuses on measuring the time lag behavior of the voltage response. Within each perturbation time window, the system first determines the time when the micro load starts to apply according to the current signal, which is usually represented by the time point when the current jumps from the stable state close to zero to the specified micro-perturbation current value. The system then analyzes the rate of change of the corresponding voltage sequence, i.e., by dividing the voltage difference between consecutive time points by the time interval, to construct a voltage rate of change sequence. The time point corresponding to the maximum rate of change in this sequence is found, i.e., the time when the voltage changes the fastest, which is usually the response point when the voltage starts to drop or rise rapidly. The system takes the time interval between this time point and the start of the current perturbation as the delay value of the voltage response. To maintain comparability at different time window scales, the delay value is also divided by the total length of the current time window, converting it into a dimensionless relative value, i.e., the delay response index. This index is usually between 0 and 1, and the larger the value, the slower the battery's response to the perturbation, which may represent physical phenomena such as the increase in diffusion impedance or the delay of interface layer charge response.

[0060] The extraction process of these two characteristic quantities can be automatically implemented on a digital signal processing platform, independent of manual judgment, suitable for standardized batch evaluation procedures, and can be integrated into battery test systems, controllers or embedded analysis platforms through programming.

[0061] Furthermore, the battery internal resistance dynamic evolution tensor is applied to a multi-scale time window segmentation strategy for local feature extraction to obtain a feature vector group containing a voltage nonlinear rebound factor and a delay response index, including: According to the difference between the maximum rebound speed of the recovery segment corresponding to each segment in the battery internal resistance dynamic evolution tensor and the delay take-off time, a local perturbation behavior index table is constructed to mark the relative time positions of the voltage change most severe segment and the response take-off delay segment in the tensor; Under the guidance of the local perturbation behavior index table, a rebound speed peak segment is taken as the center, and a plurality of time windows with different lengths are applied for sliding segmentation, the time window lengths include at least 50 milliseconds, 100 milliseconds, 200 milliseconds and 500 milliseconds, and each time window coverage area needs to contain at least one marked position; In each time window covered by the tensor recovery segment, a first-order linear trend fitting is performed, and the residual sum of squares between the actual voltage sampling value in the time window and the fitted trend line is normalized as the voltage nonlinear rebound factor corresponding to the time window; The time difference between the perturbation current transition time and the time point corresponding to the maximum voltage change rate in the tensor is calculated in each time window, and the time difference is normalized as the delay response index corresponding to the time window; The voltage nonlinear rebound factor calculated in all time windows and the delay response index are combined into feature pairs, respectively, and all feature pairs are organized into the feature vector group according to the disturbance segment index and the time window sequence as the disturbance behavior input feature of the subsequent input of the trained Bayesian weighted adaptive mapping network.

[0062] In the battery self-discharge characteristic analysis and rapid evaluation method described in the application, in order to effectively extract representative disturbance behavior features from the battery internal resistance dynamic evolution tensor to support subsequent accurate modeling and classification of the battery self-discharge rate, the system further performs multi-scale time window segmentation and local physical feature extraction operations on the basis of constructing the tensor, and finally generates a set of feature vector groups with engineering interpretation and model usability. This process not only combines the dynamic structure of the multi-segment disturbance response in the tensor, but also introduces a calibrated reference index closely related to the disturbance intensity and response hysteresis, ensuring the focus and discriminability of local feature extraction.

[0063] First, in the constructed battery internal resistance dynamic evolution tensor, the system extracts two important behavior features in the recovery segment of each disturbance segment: one is the maximum rebound speed, which represents the steepest recovery rate of the voltage after the disturbance is removed; the second is the delay take-off time difference, which represents the time delay from the end of the disturbance current to the obvious recovery of the voltage. These two physical quantities reflect the speed of battery polarization release and diffusion reaction, and have high individualized behavior description ability. The system takes these two features as the core to construct a local disturbance behavior index table to record the position of the voltage change most severe section and the take-off delay area in the tensor. The index table provides a structured positioning basis for subsequent guidance of the starting point selection and scale adaptation of the sliding time window.

[0064] Based on the above disturbance behavior index table, the system performs multi-scale sliding time window segmentation operation within each disturbance segment with the time point corresponding to the maximum rebound speed as the center. The selected time window length covers multiple typical scales, including 50 milliseconds, 100 milliseconds, 200 milliseconds and 500 milliseconds, covering the fast recovery and slow diffusion behavior features commonly seen in battery dynamic response. In order to ensure that the segmented window can capture the most critical physical behavior features, the tensor region covered by each time window must contain at least one position marked in the index table, that is, to ensure that the time window center and the behavior inflection point are aligned.

[0065] In each segmented time window, the system first extracts the complete sample sequence of the voltage recovery segment and performs a first-order linear trend fitting on the sequence. The fitting method can use the least squares linear fitting method to construct an equal-interval slope baseline on the time axis to simulate the ideal linear rebound path. Then, the system squares the point-to-point residual between the actual voltage sample values in the time window and the fitted trend line, and then normalizes the residual sum of squares by the number of sample points and the voltage amplitude range to obtain a dimensionless index, i.e. the voltage nonlinear rebound factor corresponding to the time window. The factor reflects the deviation between the voltage recovery process and the ideal linear model, and the larger the value, the more irregular or intense the recovery process, which is common in high-impedance degraded batteries or severely polarized single cells.

[0066] At the same time, the system also extracts another key indicator in each time window: the delay response index. The index is calculated based on the time difference between the disturbance current transition time and the time point corresponding to the maximum voltage change rate in the tensor. Specifically, the system first identifies the starting time of the perturbation by the current signal, and locates the time in the tensor; then analyzes the first-order difference of the voltage sequence and finds the sampling time point corresponding to the maximum change rate. The time interval between the two is the delay time, and the time difference is divided by the current time window length to convert it to a proportional form, forming the delay response index. The index characterizes the lag of the voltage response relative to the current disturbance and is an important representation of the dynamic response speed.

[0067] Finally, the system pairs and combines the voltage nonlinear rebound factor and the delay response index calculated in each time window to form a set of feature pairs. After processing all the time windows of all the disturbance segments in the entire tensor, the system organizes all the feature pairs according to the segment index and the time window order to form the final feature vector set. The feature vector set not only maintains the time local structure and physical change logic of the disturbance behavior, but also uniformly encodes the response patterns at different scales, with good model input adaptability.

[0068] The above feature vector set will be used as the input feature of the subsequent input of the trained Bayesian weighted adaptive mapping network, supporting the system to generate battery self-discharge rate estimates with high precision and high confidence under the consideration of manufacturing batch residual modeling and temperature offset compensation mechanism, providing reliable basis for structured assessment and classification.

[0069] Step S104: inputting the feature vector set into the trained Bayesian weighted adaptive mapping network to obtain the battery self-discharge rate estimate considering the manufacturing batch residual modeling and temperature offset compensation mechanism.

[0070] Further, the Bayesian weighted adaptive mapping network comprises a feature encoding unit, a manufacturing batch and temperature condition joint modeling unit, and an output fusion and confidence inference unit. The feature encoding unit is configured to receive the normalized feature vector group comprising a plurality of voltage nonlinear springback factors and delay response exponents, perform feature extraction and compression on the feature vector group through a multi-layer feedforward network, and output a fixed-length encoding vector representing the disturbance response behavior characteristics. The manufacturing batch and temperature condition joint modeling unit is configured to receive the manufacturing batch and environmental temperature information of the target battery, and splice the information with the fixed-length encoding vector to generate a joint input vector; and comprises a plurality of sub-mapping paths, each of which is optimized based on the training data of a specific manufacturing batch and temperature interval, and assigns a Bayesian posterior weight between the current input condition and the center condition vector of each sub-path during inference. The output fusion and confidence inference unit is configured to receive candidate battery self-discharge rate estimates output by all sub-mapping paths, and perform weighted fusion on each candidate value combined with the Bayesian posterior weight to generate a final battery self-discharge rate estimate; and calculate a confidence score based on the dispersion degree between the candidate values to represent the reliability of the final estimate.

[0071] In step S104, the feature vector group obtained in step S103 is input into the trained Bayesian weighted adaptive mapping network to obtain a battery self-discharge rate estimate considering manufacturing batch residual modeling and temperature offset compensation mechanism. The Bayesian weighted adaptive mapping network comprises a feature encoding unit, a manufacturing batch and temperature condition joint modeling unit, and an output fusion and confidence inference unit, which cooperatively constitute a complete mapping prediction path.

[0072] First, the system receives a normalized feature vector group composed of a plurality of voltage nonlinear springback factors and delay response index groups extracted at different time window scales. The feature vector group is input into the feature encoding unit, which is configured to map and compress the original input to generate a feature expression vector with stronger discrimination ability as the basis for subsequent modeling steps. The feature encoding unit can be implemented in a multi-layer feedforward network, a fully connected structure, or other linear / nonlinear mapping methods, and its output is a fixed-length feature encoding vector representing the disturbance response behavior characteristics of the target battery.

[0073] Subsequently, the system reads the manufacturing batch information and the test ambient temperature information of the current target battery, and encodes them into a manufacturing batch and temperature condition vector, where the manufacturing batch can be represented in one-hot encoding form, and the temperature information can be normalized. The system splices the manufacturing batch and temperature condition vector with the aforementioned feature encoding vector to form the joint input of the model.

[0074] In the manufacturing batch and temperature condition joint modeling unit, the system aims to improve the stability and adaptability of the overall prediction by modeling and compensating for the interference of manufacturing source differences and ambient temperature on the prediction results of the battery self-discharge rate. The key technical structure of this unit is a set composed of several sub-mapping paths, each corresponding to a sample distribution feature represented by a combination of manufacturing batch and temperature interval. All sub-mapping paths have been separately optimized for overfitting using sample data of the corresponding category during the training phase, thereby possessing targeted modeling capabilities.

[0075] In actual inference process, the system first encodes the manufacturing batch and temperature condition vector of the current target battery. The manufacturing batch information is derived from production labels or test databases, indicating the production batch number to which the battery belongs. The system uses one-hot encoding to convert it into an equal-length vector representation, such as "Batch 1" represented as [1, 0, 0, 0], "Batch 2" represented as [0, 1, 0, 0], etc. The temperature information is obtained through measurement systems, with units in Celsius. In order to be compatible with the model, the system normalizes the current temperature using the historical temperature range recorded in the training data set, converting it to a real value between 0 and 1. This processing ensures that inputs under different temperature conditions do not cause the model to become unbalanced due to value differences.

[0076] The system splices the above two variables to form a unified manufacturing batch and temperature condition vector. Subsequently, the system calculates the similarity between each path and the input based on this vector in the set of sub-mapping paths that have been trained, and then dynamically allocates the participation weight of each path. Each sub-mapping path records the manufacturing batch distribution and temperature statistical features it covers during training, such as mean, variance, etc. The current input condition vector will be matched with the historical distribution of each sub-path, for example, by calculating the Euclidean distance, cosine similarity, Mahalanobis distance, or Gaussian kernel function output matching score. These matching scores are normalized and used as the Bayesian posterior weight estimation value of the current input on each path.

[0077] For example, if the current battery comes from "Batch 3", the ambient temperature is 33°C, and the training data of sub-path A mainly comes from "Batch 3" at 30-35°C, which has a highly coincident distribution with the current input condition, the system will assign a higher weight to sub-path A; while sub-path B mainly comes from "Batch 1" and the training temperature is around 20°C, the system will assign it a lower weight or even close to zero. This weight assignment mechanism essentially realizes the modeling of the manufacturing batch residual of the input sample, that is, the prediction is based on the data sub-distribution from the manufacturing background with similar manufacturing background, and at the same time, the temperature is compensated by the temperature, to ensure that the prediction result will not be misjudged due to cross-environment data.

[0078] After the weight assignment is completed, the system inputs the current input feature vector group into all sub-mapping paths. Each sub-path structure is usually a group of small feedforward networks or regressors, which have been optimized using samples under the corresponding batch and temperature conditions in the training stage, and the structure remains unified but the weight parameters are different. Each sub-path outputs a candidate battery self-discharge rate estimate, which is in units of volts per hour, millivolts per day, or other custom units, as configured by the system.

[0079] All candidate battery self-discharge rate estimates are sequentially input into the output fusion and confidence reasoning unit. In this unit, the system first obtains the Bayesian weight corresponding to each candidate value, which comes from the weight assignment result of each sub-mapping path in the manufacturing batch and temperature condition joint modeling unit, which has been normalized to a probability form with a total sum of 1.

[0080] The system pairs all candidate battery self-discharge rate estimates with their corresponding Bayesian weights, and then performs a weighted fusion operation, that is, each candidate value is multiplied by its corresponding Bayesian weight, and all results are accumulated and summed. The value obtained by this operation is the final battery self-discharge rate estimate of the current target battery. The physical meaning of this value is: after considering the disturbance response behavior characteristics, manufacturing batch source, and test environment temperature conditions, the system infers the voltage natural decay rate per unit time, which can be in units of "millivolts per day", "volts per hour", etc., as set by the system calibration.

[0081] In addition to outputting the final result, the output fusion and confidence inference unit also needs to quantify the degree of confidence in the current prediction. To do this, the system treats the set of battery self-discharge rate estimates output by all sub-mapping paths participating in this round of fusion as a set of numerical samples. The system calculates the degree of dispersion of this set, which can be achieved through conventional statistical indicators such as standard deviation, maximum difference, or distribution variance. Among them, if the estimates of all sub-paths are close to consistent, i.e., the degree of dispersion between candidate values is small, the system will consider that the current input is within the training data support range, and the model prediction has high consistency and stability, thus giving a higher confidence score. Conversely, if the prediction results of each sub-path differ greatly, it indicates that the current sample may have out-of-distribution characteristics, uncovered batches, or extreme temperature conditions, and the system will reduce the confidence score.

[0082] The confidence score takes on a real value between 0 and 1. The system can set confidence level thresholds according to actual business needs for risk warning or manual review. Ultimately, the battery self-discharge rate estimate and its corresponding confidence score will be provided together as structured result data to the subsequent structured detection report generation module for report output and evaluation visualization. This mechanism ensures that the system not only has numerical prediction capabilities but also provides a criterion for prediction reliability, which helps engineers make quick decisions in quality control or abnormal screening scenarios.

[0083] Through the above structure, the system can output battery self-discharge rate estimates with manufacturing batch residual modeling and temperature offset compensation mechanisms based on the feature vector set of perturbation response behavior and the known manufacturing batch and temperature condition vectors. This method is suitable for rapid evaluation of batteries from different sources in various test environments and can effectively improve the accuracy, stability, and industrial applicability of predictions.

[0084] Here is the reference implementation code for the Bayesian weighted adaptive mapping network: import numpy as np # Input data structure definition # Input feature vector set: e.g. 20 dimensions (10 voltage non-linear rebound factors + 10 delay response exponents) feature_vector = np.random.rand(20) # Simulate perturbation feature vector set (normalized) # Current battery manufacturing batch (assuming 4 batches, using one-hot encoding) batch_id = 2 batch_one_hot = np.eye(4)[batch_id] # [0, 0, 1, 0] # Current battery's ambient temperature (in Celsius, normalized to [0, 1], assuming temperature range is [20, 40] Celsius) temperature_celsius = 33.0 temperature_norm = (temperature_celsius - 20.0) / (40.0 - 20.0) # Normalize to [0,1] # Concatenate the manufacturing batch and temperature condition vectors condition_vector = np.concatenate([batch_one_hot, [temperature_norm]]) # Feature encoding unit def encode_features(input_vector): # Multi-layer feedforward structure (can be considered as part of a shallow neural network) to simulate extracting behavioral feature embeddings # Using linear mapping with ReLU activation here W1 = np.random.randn(32, len(input_vector)) 0.1 b1 = np.zeros(32) hidden = np.dot(W1, input_vector) + b1 encoded = np.maximum(hidden, 0) # ReLU activation return encoded # Output a 32-dimensional encoded vector encoded_vector = encode_features(feature_vector) # Concatenate the feature encoded vector and condition vector as joint input joint_input = np.concatenate([encoded_vector, condition_vector]) # Sub-mapping path definitions (assuming there are N sub-paths in the system) # Each sub-path is optimized for a certain batch-temperature interval and records the center condition vector N_PATHS = 5 subpaths = [] for i in range(N_PATHS): path = { "center_condition": np.random.rand(len(condition_vector)), # Batch temperature distribution center of the subpath "model_weights": np.random.randn(1, len(encoded_vector)), # Simulate sub-model weights "model_bias": np.random.randn(1)[0] # Bias } subpaths.append(path) # Calculate similarity & Bayesian posterior weight assignment (using cosine similarity + softmax normalization) def compute_cosine_similarity(a, b): return np.dot(a, b) / (np.linalg.norm(a) np.linalg.norm(b) +1e-8) similarities = [] for path in subpaths: similarity = compute_cosine_similarity(condition_vector, path["center_condition"]) similarities.append(similarity) # Softmax normalization yields Bayesian posterior weights (ensuring the sum is 1). exp_sim = np.exp(similarities - np.max(similarities)) bayesian_weights = exp_sim / np.sum(exp_sim) # Output candidate self-discharge rate estimates for each sub-path candidate_outputs = [] for i, path in enumerate(subpaths): Linear regression output using fixed weights in sub-paths (a real trained regressor can also be used) weights = path["model_weights"] bias = path["model_bias"] candidate_value = float(np.dot(weights, encoded_vector) + bias) candidate_outputs.append(candidate_value) Output fusion and confidence inference unit, weighted fusion to get the final estimate final_estimate = sum(w v for w, v in zip(bayesian_weights,candidate_outputs)) Calculate confidence (use inverse standard deviation logic) std_dev = np.std(candidate_outputs) confidence_score = np.exp(-std_dev) # The smaller the dispersion, the higher the confidence (between 0 and 1) The training process of the Bayesian weighted adaptive mapping network includes three stages. First, the system divides the samples into several subsets according to the disturbance response feature vector group of the battery in the historical data, the manufacturing batch label and the test temperature, each subset corresponds to a manufacturing batch and temperature interval combination, and establishes a sub-mapping path for each subset. Second, the system trains each sub-mapping path independently, so that it can predict the battery self-discharge rate under the corresponding conditions from the input feature vector. The training process uses a standard regression loss function such as mean square error, and cross-validation is used to prevent overfitting. Finally, the system calculates the average of the condition vector in each sub-path in the training sample as the condition center of the path, which is used for similarity matching and Bayesian weight allocation in the inference stage, so as to realize the distribution awareness and condition adaptation ability. This training mechanism not only retains the pertinence of local modeling, but also supports the dynamic fusion of input conditions, ensuring stable prediction performance of the model under multi-batch and multi-environment data.

[0085] Step S105: Trend curve fitting is performed on the battery self-discharge rate estimate values of at least three consecutive periods, and the consistency is tested to obtain the battery self-discharge level evaluation result.

[0086] In step S105, the system performs time trend analysis and consistency judgment on multiple estimation results of the same target battery collected at different time periods based on the battery self-discharge rate estimation value obtained in step S104, thereby obtaining a battery self-discharge level evaluation result for level classification. To ensure the reliability of the evaluation result, this step requires at least three or more self-discharge rate estimation values in independent time segments, which can be obtained under different disturbance periods or different sampling windows, provided that they correspond to the same physical battery and the test environment remains consistent.

[0087] In a specific implementation process, the system first time-labels all battery self-discharge rate estimation values of the same target battery, records the test order when each value is generated, and organizes them into an estimation value sequence in chronological order. Each value has complete input features, model structure, manufacturing batch, and temperature conditions when it is generated, so the system can confirm the internal consistency of the sequence and determine whether it can be used for subsequent trend fitting processing. If the number of samples in the sequence is less than three, the system will abort this step and return a "data insufficient" identifier to avoid level judgment errors.

[0088] After the estimation value sequence meets the minimum sample number, the system takes this group of estimation values as input and performs trend fitting. Trend fitting refers to numerical analysis of multiple self-discharge rate estimation values by the system to identify whether their change trend is stable, such as whether it is continuously rising, falling, or fluctuating sharply within a short period of time. The system can use methods such as moving average or piecewise linear fitting to model the overall trend of the estimation value sequence. The system analyzes the trend by judging the following three indicators: First, the system analyzes the maximum difference between all estimation values, i.e., finds the absolute difference between the highest and lowest estimation values. If the difference is less than the system's set absolute tolerance (e.g., 1 mV per day), the change range of the values is acceptable.

[0089] Second, the system judges whether the change direction between the estimation values is consistent, i.e., whether the change between adjacent values is increasing, decreasing, or fluctuating. For example, if the values show a rising-then-falling pattern, they can be considered to have high volatility; if the change always shows a decreasing trend, it means that the self-discharge rate tends to weaken, which may be a recovery-type battery.

[0090] Third, the system evaluates the relative standard deviation between multiple estimation values, i.e., the ratio of the magnitude of the deviation of multiple values from the mean to the mean itself, to quantify volatility. The system sets a threshold for this indicator, e.g., allowing the standard deviation to be no more than 15% of the mean.

[0091] When the above three indicators meet the system's preset stability conditions, the system determines that the self-discharge rate estimate of the target battery has good consistency and has a statistical basis for grade evaluation. Subsequently, the system grades the battery according to the average estimate. The grading standard can be preset by the user, for example, daily self-discharge rate less than 1 mV is classified as first-class excellent, 1-2 mV is classified as second-class qualified, and more than 2 mV is classified as third-class risk. The grading logic should be consistent with the product quality control requirements and can be adjusted through the system parameter configuration interface.

[0092] If the trend fitting shows that the estimates have obvious inconsistency, for example, the difference between multiple values exceeds a set threshold, or there is a jump trend, non-monotonic change, the system will output an "estimate inconsistency" prompt and refuse to output the grade result, prompting the user to re-collect data or review the model input settings.

[0093] Finally, the battery self-discharge grade evaluation result output by this step is a standardized label, such as "first-class", "second-class", "third-class" or "result invalid" classification results, which can be directly used for structured detection report summary, judgment or display, and also provides input basis for step S106.

[0094] Further, the trend curve fitting of the battery self-discharge rate estimate values of the consecutive at least three periods is performed, and the consistency is tested to obtain the battery self-discharge grade evaluation result, including: The battery self-discharge rate estimate values of the consecutive at least three periods are constructed into an ordered estimate sequence in time sequence, and the rebound amplitude corresponding to each period is recorded to form a paired sequence of self-discharge rate and rebound amplitude; The difference between the self-discharge rate estimate values of any two adjacent periods in the estimate sequence is calculated, and the change direction is determined, and the number of times of change direction reversal in the entire sequence is further counted, and if the number of times of reversal is greater than or equal to two, the trend fluctuation flag of the sequence is output; According to the trend fluctuation flag, the paired sequence is traversed, the period with the maximum rebound amplitude increase is extracted, and the change amount of the self-discharge rate estimate value of the period relative to the previous period is calculated as the offset sensitive difference value; When the trend fluctuation flag exists and the offset sensitive difference value is greater than a preset offset expansion threshold, a judgment result that the consistency condition is not met is output, and the operation of obtaining the instantaneous voltage change sequence and the current micro-perturbation response sequence of the target battery under micro-load disturbance is triggered; otherwise, the ordered estimate sequence is output as the basis for generating the battery self-discharge grade evaluation result.

[0095] In the battery self-discharge characteristic analysis and rapid evaluation method described in the application, in order to improve the accuracy of judging the battery self-discharge behavior and the stability of the results, the change trend of the battery self-discharge rate estimate value in multiple periods is analyzed, and its consistency is tested, so as to judge whether the battery self-discharge grade evaluation result can be directly generated according to the above or whether the data needs to be re-collected to perform re-evaluation process.

[0096] Firstly, the battery self-discharge rate estimate values obtained in at least three consecutive periods are arranged in time sequence one by one to form an ordered estimate sequence. In order to introduce the coupling constraint of the change of the physical state of the battery, the rebound amplitude of each period is also extracted synchronously, and is paired with the battery self-discharge rate estimate value of the corresponding period one by one to form a paired data structure containing behavior characteristics. The structure is used for subsequent judgment of whether the change of the self-discharge rate is affected by the fluctuation of the rebound behavior, and provides a basis for trend identification and deviation analysis.

[0097] Subsequently, the above estimate sequence is subjected to difference analysis, that is, the difference between the battery self-discharge rate estimate values of any two adjacent periods is calculated, and the change direction is judged according to the difference. By counting the change direction of consecutive periods, if it is found that the rate change direction has been reversed twice or more, that is, the conversion from rising to falling or from falling to rising, it is considered that the sequence has unstable trend behavior. At this time, a trend fluctuation flag is generated to identify the repeated shock or unexplained reverse fluctuation of the overall estimate trend of the sequence, prompting the possible existence of local disturbance, measurement error or external temperature disturbance risk.

[0098] After the trend fluctuation flag is generated, the paired sequence is continued to be traversed based on the change of the rebound amplitude. The rebound amplitude values in all periods are compared, the period with the most significant rebound amplitude growth is extracted, and the numerical change amount between the battery self-discharge rate estimate value at this period and the estimate value of the previous period is calculated with the period as a reference to form a deviation sensitive difference value. The difference value is used to evaluate whether the self-discharge rate estimate value also has a significant synchronous rise under the background of the sharp enhancement of the rebound behavior, so as to judge whether the estimate value truly reflects the change of the internal state of the battery or is a local abnormal amplification under the interference of external factors.

[0099] Finally, a consistency determination is performed according to the combination of the trend fluctuation flag and the offset sensitive difference. If the trend fluctuation flag exists in the estimated value sequence, and the offset sensitive difference exceeds the preset offset expansion threshold, it means that the estimated value sequence does not meet the consistency requirements in terms of behavior continuity and amplitude stability, and cannot be directly used for evaluation. Therefore, the system should reacquire the instantaneous voltage change sequence and the current micro-disturbance response sequence of the target battery under micro-load disturbance in multiple short time segments, reextract the features and estimate the self-discharge rate. On the contrary, if the above two conditions do not simultaneously exist, the estimated value sequence is considered to have an acceptable consistency level, and can be directly used as the basis for generating the battery self-discharge grade evaluation result, for subsequent report generation and quality classification process.

[0100] The consistency verification strategy is based on the structural analysis and physical behavior coupled judgment of multi-cycle features, does not rely on machine learning models or complex statistical curves, but is completed through the coordination of direction variation, amplitude jump and judgment threshold of limited data points, is suitable for large-scale and periodic battery evaluation scenarios, and can be realized by logic control at the software layer of the existing battery test platform.

[0101] Step S106: Based on the battery self-discharge grade evaluation result and the corresponding battery self-discharge rate estimate value, a structured detection report containing the initial offset, the rebound amplitude and the estimation confidence is generated.

[0102] In step S106, the system generates a detection report with complete structured format based on the battery self-discharge grade evaluation result and the corresponding battery self-discharge rate estimate value output in step S105. The detection report contains at least the initial offset, the rebound amplitude and the estimation confidence. The report is used to support battery quality management, factory screening and subsequent process optimization, and its structure and content should ensure that it can be automatically read, archived and analyzed.

[0103] Firstly, the system extracts each group of disturbance response original segments used to generate the battery self-discharge rate estimate value, and traces back to the instantaneous voltage change sequence corresponding to each estimate value. In these sequences, the system identifies the voltage minimum point after the disturbance ends and the subsequent voltage recovery process, and further extracts two core physical behavior parameters, namely the initial offset and the rebound amplitude, by analyzing the voltage change trend in the recovery process.

[0104] The initial offset refers to the voltage drop amplitude at the moment of disturbance end relative to the stable voltage before disturbance, and the unit is volt. This indicator is used to reflect the instantaneous voltage drop degree of the battery under micro-load disturbance. The larger the value is, the more significant the internal resistance or polarization effect may be. The system calculates this indicator by extracting the static voltage difference between the disturbance excitation segment and the disturbance end moment.

[0105] Rebound amplitude refers to the voltage uplift between the lowest point and the final stable value during the recovery process after the disturbance ends, also in volts. This indicator reflects the battery polarization relief capability and the recovery efficiency of the internal diffusion mechanism, and is usually used as an additional reference to evaluate the battery response lag.

[0106] Next, the system will call the confidence score output in step S104 as the estimated confidence in this report, which is generated by the output fusion and confidence inference unit according to the dispersion degree between all sub-mapping path output results, to reflect the credibility of the current prediction result. The confidence score is a standardized score ranging from 0 to 1, and the system can set a threshold value to identify whether manual review is needed.

[0107] The report structure uses standardized data fields to organize each report, which includes the following information fields: unique identification number of the target battery, sampling time, test environment temperature, manufacturing batch identification, all self-discharge rate estimation value sequences, final self-discharge rate estimation value, corresponding confidence score, battery self-discharge grade evaluation result generated by the system according to the preset rules, initial offset, voltage rebound amplitude, whether to trigger a review suggestion, etc. The report can be exported as JSON, CSV or database structured records, which is convenient for calling in automated production lines, battery sorting systems or experimental data platforms.

[0108] In addition, the system supports further sorting, filtering and alarming based on the structured detection report. For example, if the battery is evaluated as a "third" self-discharge grade and its confidence is lower than the set threshold, the system can insert a prompt label "suggest retesting" in the report and mark it as needing manual review. If all indicators perform well, the system can automatically identify it as "qualified" and mark the corresponding grade.

[0104] Finally, the generated structured detection report not only reflects the quantitative estimation value output, but also supplements the qualitative evaluation dimension of voltage response behavior in an engineering way, with good traceability, interpretability and batch scalability, suitable for large-scale deployment and operation of battery manufacturing enterprises, quality inspection agencies and backend system integrators.

[0110] Further, based on the battery self-discharge grade evaluation result and the corresponding battery self-discharge rate estimation value, a structured detection report containing the initial offset, the rebound amplitude and the estimation confidence is generated, including: Based on the battery self-discharge level evaluation result and the corresponding battery self-discharge rate estimation value, the starting offset, the rebound amplitude and the estimation confidence of the current period are extracted, and are sequentially arranged according to the calculation order of the Bayesian weighted adaptive mapping network in the generation of the battery self-discharge rate estimation value, so as to construct a detection field list corresponding to the structural field order and the calculation path logic; According to the disturbance segment index of the corresponding feature vector group in the detection field list, the data segment to which the corresponding original instantaneous voltage change sequence and the current micro-perturbation response sequence belong is determined, and based on the disturbance segment index, the time window number and the recovery segment starting voltage, a disturbance segment encoding fingerprint is generated, which is added to the structured detection report as a unique identifier for data tracking and report tracing; Combined with the battery internal resistance dynamic evolution tensor, the forward rebound speed and the reverse voltage drop speed of each disturbance segment recovery segment are extracted, the maximum difference between the two is calculated as a period rebound asymmetry index, and the period rebound asymmetry index is added to the detection field list as a supplementary field for assisting battery quality rating; A structured detection report containing the starting offset, the rebound amplitude, the estimation confidence, the disturbance segment encoding fingerprint and the period rebound asymmetry index is generated, the structured detection report is attached with a field-level hash check value generated based on the field content, and the current data version identification is recorded, so as to support the content integrity verification and historical version comparison analysis of the report result.

[0111] In the battery self-discharge characteristic analysis and rapid evaluation method described in the application, the generation of the structured detection report is not only a result display process, but also a deep data packaging step around the self-discharge behavior extraction, path tracking and quality attribution. The detection report organizes the key calculation outputs in the foregoing evaluation process through a rigorous data structure organization form, and at the same time enhances the engineering application value, credibility and traceability of the report through the information tracing and abnormal analysis mechanism.

[0112] In this process, first, based on the obtained battery self-discharge level evaluation results and their corresponding battery self-discharge rate estimates, three key feature values are extracted from the current period: the initial offset, the rebound amplitude, and the estimation confidence. Among them, the initial offset refers to the sinking amplitude of the voltage relative to the pre-disturbance steady-state voltage at the end of the disturbance, which reflects the degree of polarization residue after discharge; the rebound amplitude describes the rising amount of the voltage from the lowest point to the steady state, which reflects the recovery ability; and the estimation confidence comes from the confidence inference result in the aforementioned Bayesian weighted adaptive mapping network, which represents the confidence level of the current self-discharge rate estimate. These three data are arranged in order, and the order is based on the derivation path order of the corresponding calculation in the Bayesian weighted adaptive mapping network during the battery self-discharge rate estimate generation process, ensuring that the field order of the detection report is consistent with the calculation logic of the evaluation model, thereby forming a detection field list with consistent structure field order and calculation path logic.

[0113] Subsequently, to enhance the traceability of the report content, the system traces the source of the original disturbance data according to the feature vector group segment index corresponding to each record in the detection field list. In the aforementioned feature vector group generation process, each vector group comes from a certain disturbance segment in the battery internal resistance dynamic evolution tensor. The system uses the index number of the disturbance segment, the time window number it is in, and the starting voltage value of the recovery segment to construct a disturbance segment encoding fingerprint. This encoding fingerprint is generated in the form of a hash function, which provides a unique identifier without revealing the original measurement data content. This fingerprint is bound and written to a fixed field position in the structured detection report, serving as the basis for subsequent verification of the accuracy of the report fields, comparison of the original data, and analysis of abnormalities.

[0114] In terms of supplementary feature dimensions, the structured detection report also includes a period rebound asymmetry index for quality rating enhancement. This index is calculated from the data of the recovery segment in the battery internal resistance dynamic evolution tensor. Specifically, the system extracts the maximum rebound speed during the voltage rise in the disturbance recovery phase, as well as the maximum negative recovery speed in the subsequent possible voltage secondary drop process, and calculates the maximum difference between the two. This difference reflects whether the recovery curve form has symmetry deviation, which can reveal features such as delayed polarization release, slow instability, or local drop, and is commonly used in engineering to identify suboptimal monomers, edge-aged batteries, or composite fault signs. This asymmetry index is added as a supplementary field to the detection field list, along with the aforementioned initial offset, rebound amplitude, and estimation confidence, to form the core index set of the structured report.

[0115] Finally, the system combines all the above fields to generate a structured detection report. This report not only contains the starting offset, the rebound amplitude, and the estimated confidence, but also contains the perturbation segment encoding fingerprint and the period rebound asymmetry index, ensuring that the structure is comprehensive, the source is clear, and the indicators are complementary. To prevent report fields from being tampered with or mismatched, the system performs field-level hash verification on all field contents, and attaches a digest signature formed by the field order, field value, and check value set corresponding to this version, while recording the version number and configuration identifier used by the current data processing logic. This not only ensures the integrity verification and engineering audit of each structured detection report content, but also facilitates subsequent algorithm iteration, data batch comparison, and detection platform version backtracking management.

[0116] The structured detection report generated in the above manner has visual friendliness, data verification capability, content credibility, and behavior explanation path, which can significantly improve the efficiency and standardization level of large-scale automatic detection and grading quality analysis of batteries.

[0117] The second embodiment of the application provides an electronic device, which comprises: a processor; a memory for storing a program, which, when read and executed by the processor, executes the battery self-discharge characteristic analysis and rapid evaluation method provided in the first embodiment of the application.

[0118] The third embodiment of the application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, executes the battery self-discharge characteristic analysis and rapid evaluation method provided in the first embodiment of the application.

[0119] Although the above is disclosed in the preferred embodiments of the application, it is not intended to limit the application, and any person skilled in the art can make possible changes and modifications without departing from the spirit and scope of the application. Therefore, the protection scope of the application should be subject to the scope defined by the claims of the application.

Claims

1. A method for analyzing and rapidly evaluating the self-discharge characteristics of a battery, characterized in that, include: Within a set environmental stability range, acquire the instantaneous voltage change sequence and current perturbation response sequence of multiple short time segments of the target battery under micro-load perturbation; Based on the instantaneous voltage change sequence and current perturbation response sequence, a battery internal resistance dynamic evolution tensor is constructed to represent the dynamic response characteristics of the battery. For the battery internal resistance dynamic evolution tensor, a multi-scale time window segmentation strategy is applied to extract local features, and a feature vector set containing voltage nonlinear rebound factor and delay response exponent is obtained. The feature vector set is input into the trained Bayesian weighted adaptive mapping network to obtain the battery self-discharge rate estimate considering manufacturing batch residual modeling and temperature offset compensation mechanism. The estimated self-discharge rate of the battery for at least three consecutive cycles is fitted with a trend curve, and the consistency of the curve is checked to obtain the battery self-discharge level assessment result. Based on the battery self-discharge level assessment results and the corresponding battery self-discharge rate estimate, a structured test report is generated, which includes the initial offset, rebound amplitude, and estimated confidence level.

2. The method for analyzing and rapidly evaluating battery self-discharge characteristics according to claim 1, characterized in that, The micro-load disturbance refers to applying a short-time pulse current signal with a current amplitude less than 1% of the rated capacity to the target battery during the test, and the duration of each disturbance is less than 2 seconds, so as to avoid significantly changing the battery's state of charge.

3. The method for analyzing and rapidly evaluating battery self-discharge characteristics according to claim 1, characterized in that, The voltage nonlinear rebound factor represents the degree of deviation of the voltage recovery process from the linear trend after the disturbance ends. It is obtained by fitting the voltage sequence within the disturbance recovery segment and calculating the average squared residual. The delay response index is used to reflect the voltage response time delay after the current disturbance is applied. Its value is the time interval between the current transition moment and the moment of maximum voltage change, and is normalized to the proportion of the disturbance time window length.

4. The method for analyzing and rapidly evaluating battery self-discharge characteristics according to claim 1, characterized in that, The Bayesian weighted adaptive mapping network includes a feature encoding unit, a joint modeling unit for manufacturing batches and temperature conditions, and an output fusion and confidence inference unit. The feature encoding unit is used to receive a normalized feature vector group, which includes multiple voltage nonlinear rebound factors and delay response exponents. The feature vector group is processed by a multi-layer feedforward network to extract and compress features, and output a fixed-length encoding vector to represent the characteristics of disturbance response behavior. The manufacturing batch and temperature condition joint modeling unit is used to receive the manufacturing batch and ambient temperature information of the target battery, and concatenate it with the fixed-length encoding vector to generate a joint input vector; it includes multiple sub-mapping paths, each of which is optimized based on training data of a specific manufacturing batch and temperature range, and Bayesian posterior weights are assigned during inference based on the similarity between the current input conditions and the central condition vector of each sub-path. The output fusion and confidence inference unit is used to receive the candidate battery self-discharge rate estimates output by all sub-mapping paths, and to perform weighted fusion of each candidate value in combination with the Bayesian posterior weight to generate the final battery self-discharge rate estimate; and to calculate the confidence score based on the degree of dispersion between candidate values ​​to characterize the reliability of the final battery self-discharge rate estimate.

5. The method for analyzing and rapidly evaluating battery self-discharge characteristics according to claim 1, characterized in that, The construction of a battery internal resistance dynamic evolution tensor to represent the battery's internal dynamic response characteristics, based on the instantaneous voltage change sequence and current perturbation response sequence, includes: Within the set environmental stability range, the instantaneous voltage change sequence and current perturbation response sequence of each group of short time segments are divided into a pre-perturbation resting segment, a perturbation excitation segment, and a perturbation end recovery segment, respectively. The segment division is based on the initial slope and amplitude threshold of the current change, and the boundary time is automatically determined. In each perturbation excitation segment, the time relationship between the voltage drop start point, the voltage minimum point and the current transition point corresponding to the current perturbation is extracted and marked as the voltage response window index field, which serves as the time positioning reference for subsequent tensor construction. In each disturbance end recovery segment, the recovery velocity vector between continuous voltage sampling points within the segment is calculated, and combined with the initial recovery offset value, maximum rebound velocity and rebound amplitude, a three-dimensional battery internal resistance dynamic evolution tensor with time step, ternary physical characteristics and segment number is constructed. The maximum rebound velocity and the time difference between the delayed start-up time corresponding to each segment of the battery internal resistance dynamic evolution tensor are used as local dynamic feature fields to participate in the subsequent operation of local feature extraction using a multi-scale time window segmentation strategy.

6. The method for analyzing and rapidly evaluating battery self-discharge characteristics according to claim 5, characterized in that, The dynamic evolution tensor of the battery internal resistance is subjected to a multi-scale time window segmentation strategy for local feature extraction, resulting in a feature vector set containing the voltage nonlinear rebound factor and the delay response exponent, including: Based on the difference between the maximum rebound speed of the recovery segment and the delayed start-up time for each segment in the dynamic evolution tensor of the battery internal resistance, a local perturbation behavior index table is constructed to mark the relative time positions of the most drastic voltage change segment and the response start-up hysteresis segment in the tensor. Guided by the local disturbance behavior index table, multiple time windows of different lengths are applied to slide and divide the area centered on the peak rebound speed segment. The time window lengths include at least 50 milliseconds, 100 milliseconds, 200 milliseconds, and 500 milliseconds, and each time window coverage area must include at least one marked position. In the tensor recovery segment covered by each time window, a first-order linear trend fitting is performed, and the sum of squared residuals between the actual voltage sampled value and the fitted trend line within the time window is normalized as the voltage nonlinear rebound factor corresponding to the time window. Within each time window, the time difference between the transition moment of the disturbance current and the time point corresponding to the maximum voltage change rate in the tensor is calculated, and the time difference is normalized to serve as the delay response index corresponding to the time window. The voltage nonlinear rebound factor and delay response exponent calculated in all time windows are combined into feature pairs, and all feature pairs are organized into feature vector groups according to the perturbation segment index and time window order, which are used as perturbation behavior input features for the subsequently trained Bayesian weighted adaptive mapping network.

7. The method for analyzing and rapidly evaluating battery self-discharge characteristics according to claim 1, characterized in that, The process of fitting a trend curve to the estimated battery self-discharge rate over at least three consecutive cycles and verifying its consistency to obtain a battery self-discharge level assessment result includes: The estimated self-discharge rate of the battery for at least three consecutive cycles is constructed into an ordered estimation sequence in chronological order, and the rebound amplitude corresponding to each cycle is recorded together to form a pairing sequence of self-discharge rate and rebound amplitude. The self-discharge rate estimate between any two adjacent periods in the estimated sequence is calculated by difference, and its direction of change is determined. The number of times the direction of change reverses in the entire sequence is further counted. If the number of reversals is greater than or equal to two, the trend fluctuation indicator of the sequence is output. Based on the trend fluctuation indicator, the paired sequence is traversed, the period with the largest increase in rebound amplitude is extracted, and the change in the self-discharge rate estimate of the period relative to the previous period is calculated as the offset sensitive difference. When the trend fluctuation indicator exists and the offset sensitivity difference is greater than the preset offset expansion threshold, the judgment result that does not meet the consistency condition is output, and the aforementioned operation of obtaining the instantaneous voltage change sequence and current micro-perturbation response sequence of multiple short time segments of the target battery under micro-load disturbance is triggered; otherwise, the ordered estimation sequence is output as the basis for generating the battery self-discharge level evaluation result.

8. The method for analyzing and rapidly evaluating battery self-discharge characteristics according to claim 1, characterized in that, Based on the battery self-discharge level assessment results and the corresponding estimated battery self-discharge rate, a structured inspection report is generated, including the initial offset, rebound amplitude, and estimated confidence level. Based on the battery self-discharge level assessment results and the corresponding battery self-discharge rate estimate, the starting offset, rebound amplitude and estimated confidence of the current cycle are extracted and arranged in the order of calculation when the Bayesian weighted adaptive mapping network generates the battery self-discharge rate estimate. The three are then constructed into a list of detection fields whose structural field order corresponds to the calculation path logic. Based on the disturbance segment index of the feature vector group corresponding to the detection field list, determine the data segment to which the corresponding original instantaneous voltage change sequence and current perturbation response sequence belong. Based on the disturbance segment index, time window number and recovery segment start voltage, generate a disturbance segment encoded fingerprint. This disturbance segment encoded fingerprint is attached to the structured detection report as a unique identifier for data tracking and report tracing. Combining the battery internal resistance dynamic evolution tensor, the forward rebound velocity and reverse voltage fall velocity of each disturbance segment recovery segment are extracted, and the maximum difference between the two is calculated as the time period rebound asymmetry index. This time period rebound asymmetry index is added to the detection field list as a supplementary field to assist battery quality rating. A structured detection report is generated, which includes the initial offset, rebound amplitude, estimated confidence level, perturbation segment encoded fingerprint, and time-period rebound asymmetry index. The structured detection report is accompanied by a field-level hash check value generated based on the field content and records the current data version identifier to support the content integrity verification of the report results and the comparative analysis of historical versions.

Images