Battery consistency evaluation method and system based on similar working condition clustering and screening

Abstract

The application discloses a battery consistency evaluation method and system based on similar working condition clustering and screening, and the method comprises the following steps: obtaining actual operation data and dividing the actual operation data into multiple charging segments and discharging segments according to charging and discharging states; performing working condition clustering processing on the multiple charging segments to obtain charging evaluation data sources; performing state screening processing on the multiple discharging segments to obtain discharging evaluation data sources; identifying internal resistance and open-circuit voltage parameters of each single battery cell based on the charging evaluation data sources, obtaining self-discharge characteristic parameters based on the discharging evaluation data sources, and obtaining voltage and temperature parameters based on the two types of data sources respectively; extracting consistency statistical features of each parameter of the single battery cell, aggregating the consistency statistical features according to months, and generating a multi-dimensional evaluation vector; determining the weight of each dimension, comprehensively evaluating the multi-dimensional evaluation vector, and outputting a quantitative evaluation result of battery consistency. The method effectively eliminates the interference of working condition differences and improves the accuracy of battery consistency evaluation.

Description

Technical Field

[0001] This invention relates to the field of battery management technology, and in particular to a battery consistency evaluation method and system based on clustering and screening under similar operating conditions. Background Technology

[0002] With the widespread application of electrified transportation vehicles, batteries, as the core energy storage devices, are typically composed of a large number of cells connected in series or parallel to meet the demands of high power and high voltage. These electrified transportation vehicles include, but are not limited to, electric vehicles, rail transit, electric ships, and electric aircraft, and the battery types include, but are not limited to, lithium-ion batteries, sodium-ion batteries, quasi-solid-state batteries, and all-solid-state batteries. As batteries age and their usage time increases, the performance consistency between cells within the battery pack gradually deteriorates, affecting system performance and potentially causing safety hazards.

[0003] Existing battery consistency assessment methods primarily rely on statistical analysis of external parameters such as voltage, temperature, and state of charge (SOC). However, these methods have significant limitations in practical applications: Firstly, actual operating conditions are complex and variable, and battery consistency statistics are highly sensitive to factors such as charge / discharge rate, SOC range, and temperature. Mixing operating data from different conditions introduces significant operating noise, preventing the assessment results from accurately reflecting the parameter differences within the battery cells. Secondly, existing methods are largely limited to directly measurable external parameters. For indirect parameters that more fundamentally reflect the cell's health, such as internal resistance, open circuit voltage (OCV), and self-discharge rate, which cannot be directly obtained through sensors, there is currently a lack of effective extraction and consistency quantification methods suitable for actual operating data. Therefore, there is an urgent need for a battery consistency assessment method that can eliminate the interference of complex operating conditions on assessment accuracy, integrate multi-dimensional indirect parameters such as internal resistance and OCV, and is applicable to actual operating data. Summary of the Invention

[0004] The purpose of this invention is to provide a battery consistency evaluation method and system based on clustering and screening of similar operating conditions. This method eliminates the interference of operating conditions on battery consistency evaluation by segmenting the charging and discharging process based on actual operating data and clustering operating conditions. Furthermore, it extracts indirect parameters such as internal resistance, open-circuit voltage, and self-discharge rate through online parameter identification based on equivalent circuit models. This solves the technical problems in the prior art, such as the evaluation results being greatly affected by operating condition noise and the evaluation dimensions being limited to directly measurable parameters.

[0005] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution: This invention provides a battery consistency evaluation method based on clustering and screening under similar operating conditions, comprising: Acquire the actual operating data of the battery, and divide the actual operating data into multiple charging segments and multiple discharging segments according to the charging and discharging state; The multiple charging segments are subjected to operating condition clustering processing to obtain charging evaluation data sources under similar operating conditions; and the multiple discharging segments are subjected to state filtering processing to obtain discharging evaluation data sources under similar operating conditions. Based on the charging evaluation data source and the discharging evaluation data source, the target parameter set for each individual cell is determined. The target parameter set includes voltage parameters, temperature parameters, internal resistance parameters, open circuit voltage parameters, and self-discharge characterization parameters. Consistency statistical features are extracted and aggregated from the target parameter set to generate a multidimensional evaluation vector; The weights of each feature dimension in the multidimensional evaluation vector are determined, and the multidimensional evaluation vector is comprehensively evaluated based on the determined weights to output a quantitative evaluation result of battery consistency.

[0006] In one embodiment of the present invention, the step of performing operating condition clustering processing on multiple charging segments to obtain charging evaluation data sources under similar operating conditions includes: Extract the current curve features of each charging segment, including current intensity features, current stability features, current distribution features, and average rate of current change features. Based on the characteristics of the current curve, multiple charging segments are clustered according to their operating conditions to obtain charging segments under similar operating conditions as the charging evaluation data source.

[0007] In one embodiment of the present invention, the step of performing operating condition clustering processing on multiple charging segments based on the current curve characteristics to obtain charging segments under similar operating conditions as the charging evaluation data source includes: Based on the current curve characteristics of each charging segment, a first feature vector corresponding to each charging segment is constructed; The first feature vector corresponding to each charging segment is standardized to obtain the second feature vector corresponding to each charging segment. A density-based noise-based spatial clustering algorithm is used to perform unsupervised clustering on the second feature vectors corresponding to each charging segment to obtain at least one cluster. The cluster with the largest number of samples is selected as the similar working condition cluster, and each charging segment corresponding to the similar working condition cluster is used as the charging evaluation data source.

[0008] In one embodiment of the present invention, the step of performing state screening processing on multiple discharge segments to obtain discharge assessment data sources under similar operating conditions includes: Discharge segments whose state of charge variation range covers a preset state of charge range and whose individual cell temperatures are within a preset temperature range are selected from multiple discharge segments as the discharge evaluation data source under similar operating conditions.

[0009] In one embodiment of the present invention, determining the target parameter set for each individual battery cell based on the charging evaluation data source and the discharging evaluation data source includes: Based on the charging evaluation data source, the internal resistance parameters and open-circuit voltage parameters of each individual cell in the battery pack are obtained by identifying the battery model parameters. The self-discharge characterization parameters of each individual cell are obtained based on the discharge evaluation data source. The voltage and temperature parameters of each individual battery cell are obtained based on the charging evaluation data source and the discharging evaluation data source, respectively.

[0010] In one embodiment of the present invention, the step of identifying and obtaining the internal resistance parameters and open-circuit voltage parameters of each individual cell in the battery pack based on the charging evaluation data source through battery model parameters includes: A first-order RC equivalent circuit model is constructed using the terminal voltage, current, and time data from the charging evaluation data source, and the open-circuit voltage is regarded as a constant value within a short time window when the change in the state of charge is less than a preset threshold. The terminal voltage difference equation is derived based on the first-order RC equivalent circuit model, and the internal resistance parameters and open-circuit voltage parameters of the battery pack are identified online using the recursive least squares algorithm. By utilizing the voltage divider principle of series circuits, and based on the identified internal resistance parameters and open-circuit voltage parameters of the battery pack, the internal resistance parameters and open-circuit voltage parameters of each individual cell in the battery pack are calculated.

[0011] In one embodiment of the present invention, the self-discharge characterization parameter is the open-circuit voltage decay rate; The step of obtaining the self-discharge characterization parameters of each individual cell based on the discharge evaluation data source includes: For each individual battery cell, multiple discharge segments are selected from the discharge evaluation data source, where the starting point corresponding to the individual battery cell satisfies zero current and a state of charge greater than a preset state of charge threshold. From the selected multiple discharge segments, determine the discharge segment pairs with a time interval greater than or equal to a preset time threshold, and for each discharge segment pair, use the voltage at the starting point of the two discharge segments as the open circuit voltage, and calculate the single open circuit voltage decay rate based on the ratio of the difference between the two open circuit voltages to the time interval. The open-circuit voltage decay rate of the individual cell is determined based on the statistical values ​​of the multiple single open-circuit voltage decay rates corresponding to the individual cell.

[0012] In one embodiment of the present invention, the step of extracting and aggregating consistent statistical features of the target parameter set to generate a multidimensional evaluation vector includes: Consistency statistical features are extracted from the voltage parameter, temperature parameter, internal resistance parameter, open-circuit voltage parameter, and self-discharge characterization parameter, respectively, and the consistency statistical features are aggregated to generate a multidimensional evaluation vector; the consistency statistical features include standard deviation, range, coefficient of variation, mean entropy, and maximum entropy.

[0013] In one embodiment of the present invention, determining the weights of each feature dimension in the multidimensional evaluation vector includes: Construct a fuzzy evaluation matrix for each feature dimension with respect to the preset consistency evaluation level; Based on the fuzzy evaluation matrix, calculate the information entropy and difference coefficient of each feature dimension, and determine the objective weight of each feature dimension based on the information entropy and the difference coefficient. The feature dimensions are sorted from largest to smallest according to their objective weights. The top few feature dimensions whose cumulative weight contribution rate reaches a preset ratio are selected as core indicators. The weights of the selected core indicators are then normalized to obtain the final weights of each feature dimension used for comprehensive evaluation.

[0014] The battery data acquisition module is used to acquire the actual operating data of the battery and divide the actual operating data into multiple charging segments and multiple discharging segments according to the charging and discharging state. The similar operating condition extraction module is used to perform operating condition clustering processing on multiple charging segments to obtain charging evaluation data sources under similar operating conditions; and to perform state filtering processing on multiple discharging segments to obtain discharging evaluation data sources under similar operating conditions. The parameter identification module is used to determine the target parameter set of each individual cell based on the charging evaluation data source and the discharging evaluation data source. The target parameter set includes voltage parameters, temperature parameters, internal resistance parameters, open circuit voltage parameters, and self-discharge characterization parameters. The feature aggregation module is used to extract and aggregate consistent statistical features from the target parameter set to generate a multidimensional evaluation vector. The quantitative evaluation module is used to determine the weight of each feature dimension in the multidimensional evaluation vector, and to comprehensively evaluate the multidimensional evaluation vector according to the determined weights, so as to output the quantitative evaluation result of battery consistency.

[0015] As described above, this invention provides a battery consistency evaluation method and system based on clustering and screening under similar operating conditions. The method includes: acquiring actual operating data and dividing it into multiple charging and discharging segments according to charging and discharging states; extracting current curve features of the charging segments and performing operating condition clustering to obtain a charging evaluation data source; screening discharging segments based on preset state thresholds to obtain a discharging evaluation data source; identifying the internal resistance and open-circuit voltage parameters of each individual cell based on the charging evaluation data source, obtaining self-discharge characterization parameters based on the discharging evaluation data source, and obtaining voltage and temperature parameters based on both data sources; extracting consistency statistical features from the five parameters (internal resistance, open-circuit voltage, self-discharge, voltage, and temperature) of the acquired individual cells, aggregating them monthly, and generating a multi-dimensional evaluation vector; determining the weight of each feature dimension and comprehensively evaluating the multi-dimensional evaluation vector to output a quantitative evaluation result of battery consistency. This invention extracts charging segments under similar operating conditions from actual operating data through cluster analysis and screens discharging segments under a baseline state, effectively removing the coupling interference of operating condition factors such as charge / discharge rate, SOC range, and temperature on consistency indicators, ensuring that the evaluation results truly reflect the differences in the cell's intrinsic parameters. Based on differentiated processing of charging and discharging segments, online identification and extraction of indirect parameters such as internal resistance, open-circuit voltage, and self-discharge decay rate are achieved. This overcomes the limitations of existing methods that can only evaluate directly measurable parameters such as voltage and temperature, significantly expanding the dimensions and depth of consistency assessment. The entropy weighting method is used to objectively assign weights to each feature dimension and aggregate them to generate a multi-dimensional evaluation vector, achieving a comprehensive quantitative evaluation of battery consistency. The evaluation results are more robust and have greater engineering practical value. Of course, any product implementing this invention does not necessarily need to simultaneously achieve all the advantages described above. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart illustrating a battery consistency evaluation method based on clustering and screening under similar operating conditions, provided as an exemplary embodiment of this application.

[0018] Figure 2 This is a schematic diagram of a first-order RC equivalent circuit model provided for an exemplary embodiment of this application.

[0019] Figure 3 This is a schematic diagram of the structure of a battery consistency evaluation system based on clustering and screening under similar operating conditions, provided as another exemplary embodiment of this application. Detailed Implementation

[0020] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, unless otherwise specified, the following embodiments and features described therein can be combined with each other.

[0021] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. Therefore, the drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0022] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In other embodiments, publicly known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention.

[0023] To address the issues of existing technologies where evaluation results are heavily influenced by operating condition noise and the evaluation dimensions are limited to directly measurable parameters, this invention proposes a battery consistency evaluation method and system based on clustering and screening under similar operating conditions. This method first divides actual operating data into comparable similar operating condition segments based on the characteristics of the charging segment current curve and the screening based on the SOC and temperature range of the discharging segment, thereby eliminating the impact of operating condition differences on the battery consistency evaluation results. Building upon this, for the segmented similar operating condition segments, a cell-level parameter identification method based on an equivalent circuit model is proposed to obtain the internal resistance and open-circuit voltage of each individual cell. Furthermore, a K-value characterizing the open-circuit voltage decay rate is introduced to achieve a quantitative evaluation of battery self-discharge consistency. Finally, multi-dimensional consistency statistical features are integrated, and the entropy weight method is used to allocate the weights of each feature dimension. A fuzzy comprehensive evaluation method is then used to derive a comprehensive battery consistency score.

[0024] Please see Figure 1As shown in an exemplary embodiment of this application, the battery consistency evaluation method based on clustering and screening under similar operating conditions includes the following steps: S100: Acquire the actual operating data of the battery, and divide the actual operating data into multiple charging segments and multiple discharging segments according to the charging and discharging state; S200: Perform operating condition clustering processing on multiple charging segments to obtain charging evaluation data sources under similar operating conditions; and perform state filtering processing on multiple discharging segments to obtain discharging evaluation data sources under similar operating conditions. S300: Based on the charging evaluation data source and the discharging evaluation data source, determine the target parameter set for each individual cell. The target parameter set includes voltage parameters, temperature parameters, internal resistance parameters, open-circuit voltage parameters, and self-discharge characterization parameters. S400: Perform consistency statistical feature extraction and aggregation on the target parameter set to generate a multidimensional evaluation vector; S500: Determine the weights of each feature dimension in the multidimensional evaluation vector, and perform a comprehensive evaluation of the multidimensional evaluation vector based on the determined weights to output a quantitative evaluation result of battery consistency.

[0025] The steps of the above method are described in detail below with reference to specific embodiments.

[0026] First, step S100 is executed, which involves acquiring the actual operating data of the battery and dividing the actual operating data into multiple charging segments and multiple discharging segments according to the charging and discharging state.

[0027] It should be noted that the actual operating data of the battery comes from historical operating records collected in real time by the Battery Management System (BMS) and uploaded to a cloud database or local storage. The actual operating data includes parameters such as the voltage and temperature of each individual cell, the total current of the battery pack, the total voltage of the battery pack, and the state of charge of the battery pack, with each data frame accompanied by a corresponding timestamp. In this embodiment, the acquired actual operating data is first preprocessed. The preprocessing operations include, but are not limited to, missing value filling, outlier detection and removal, etc. After the preprocessing operations, preprocessed time-series data is obtained, which includes at least a timestamp, total current, total voltage, voltage of each individual cell, temperature of each individual cell, and charge / discharge status flags. Based on the charge / discharge status flags in the preprocessed time-series data, the entire time-series data is divided into multiple time-independent charging segments and multiple time-independent discharging segments. It is worth noting that, in order to eliminate the fluctuations in consistency evaluation indicators caused by differences in operating conditions (such as current rate, ambient temperature, initial state of charge, etc.) between different operating segments, this application adopts differentiated data processing strategies for charging segments and discharging segments in subsequent steps, thereby ensuring the accuracy and reliability of the evaluation results.

[0028] Next, step S200 is performed, which involves performing operating condition clustering processing on the multiple charging segments to obtain charging evaluation data sources under similar operating conditions; and performing state filtering processing on the multiple discharging segments to obtain discharging evaluation data sources under similar operating conditions.

[0029] In an exemplary embodiment of this application, step S200, performing operating condition clustering processing on multiple charging segments to obtain charging evaluation data sources under similar operating conditions, further includes: extracting current curve features of each charging segment, the current curve features including current intensity features, current stability features, current distribution features, and average rate of current change features; performing operating condition clustering processing on multiple charging segments based on the current curve features to obtain charging segments under similar operating conditions as the charging evaluation data source.

[0030] In an exemplary embodiment of this application, performing operating condition clustering processing on multiple charging segments based on the current curve characteristics to obtain charging segments under similar operating conditions as the charging evaluation data source further includes: firstly, constructing a first feature vector corresponding to each charging segment based on the current curve characteristics of each charging segment; then, standardizing the first feature vector corresponding to each charging segment to obtain a second feature vector corresponding to each charging segment; next, using a density-based noise applied spatial clustering algorithm to perform unsupervised clustering on the second feature vector corresponding to each charging segment to obtain at least one cluster; finally, selecting the cluster with the largest number of samples as the similar operating condition cluster, and using each charging segment corresponding to the similar operating condition cluster as the charging evaluation data source.

[0031] Specifically, firstly, for each charging segment, multidimensional current curve features are extracted from its charging current curve, and a first feature vector corresponding to the charging segment is constructed based on the extracted current curve features. The current curve features include current intensity features, current stability features, current distribution features, and average current change rate features. The current intensity features include the maximum current value, minimum current value, and average current value; the current stability features include the current standard deviation and the relative current fluctuation value, where the relative current fluctuation value is defined as the ratio of the current standard deviation to the current average value; the current distribution features include the current range, median, 25th percentile, 75th percentile, skewness, and kurtosis; the average current change rate feature is defined as the average absolute rate of current change between adjacent time steps, used to characterize the severity of current fluctuations over time during charging. Specifically, for a charging segment containing N sampling points, the absolute value of the current difference between each pair of adjacent sampling points is calculated sequentially in time order to obtain N-1 instantaneous current jump amplitude values, and the average value of these N-1 instantaneous current jump amplitude values ​​is determined as the average current change rate value of the charging segment. Then, the first feature vectors corresponding to each charging segment are standardized to eliminate interference from differences in dimensions and numerical ranges between feature dimensions that affect clustering analysis, thereby obtaining the second feature vectors corresponding to each charging segment. In this embodiment, the Z-score standardization method is used to perform dimensionless processing on each feature dimension, so that each processed feature dimension satisfies the statistical distribution characteristics of mean 0 and variance 1. Subsequently, the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm is used to perform unsupervised clustering on the second feature vectors corresponding to each charging segment to obtain at least one cluster. Based on this, the cluster with the largest number of samples is selected from the at least one cluster as the similar operating condition cluster, and each charging segment belonging to the similar operating condition cluster is determined as the charging evaluation data source. Since the cluster with the largest number of samples represents the charging operating condition that appears most frequently and is most representative in historical operating data, the charging segments contained in it are used as the data basis for subsequent parameter identification and consistency evaluation.

[0032] In an exemplary embodiment of this application, step S200, which involves performing state screening processing on multiple discharge segments to obtain discharge evaluation data sources under similar operating conditions, includes: selecting discharge segments from the multiple discharge segments whose state of charge change range covers a preset state of charge interval and whose individual cell temperatures are within a preset temperature interval as the discharge evaluation data sources under similar operating conditions.

[0033] Specifically, for each discharge segment obtained in step S100, the initial state of charge (SOC) value corresponding to the start time and the final SOC value corresponding to the end time of the discharge segment are extracted, and the SOC variation range of the discharge segment is determined accordingly. It is then determined whether the SOC variation range completely covers a preset SOC interval, i.e., whether the initial SOC value is greater than or equal to the upper limit of the preset SOC interval, and whether the final SOC value is less than or equal to the lower limit of the preset SOC interval. The preset SOC interval can be pre-set according to the actual application scenario and consistency evaluation requirements of the battery. In this embodiment, the preset SOC interval is set as a discharge interval with a SOC ranging from 95% to 60%. Simultaneously, it is determined whether the temperature of the individual cell corresponding to the discharge segment falls within a preset temperature range. The preset temperature range is a pre-set normal operating temperature range. In this embodiment, the preset temperature range is set to 15°C to 60°C. The purpose of setting this temperature range is to eliminate inconsistencies caused by excessively low or high ambient temperatures on the battery's self-discharge characteristics, internal resistance characteristics, and open-circuit voltage measurement accuracy. Discharge segments that simultaneously meet the following two conditions will be selected as the discharge evaluation data source: the state of charge variation range of the discharge segment covers the preset state of charge interval, and the temperature of the individual cell corresponding to the discharge segment is within the preset temperature interval. This selection mechanism ensures that the discharge data used for subsequent extraction of self-discharge characterization parameters and voltage and temperature consistency analysis originates from similar discharge conditions with sufficient discharge depth and controllable ambient temperature, thereby eliminating the impact of differences in operating conditions on the consistency evaluation results.

[0034] Next, step S300 is executed, which involves determining the target parameter set for each individual cell based on the charging evaluation data source and the discharging evaluation data source. The target parameter set includes voltage parameters, temperature parameters, internal resistance parameters, open-circuit voltage parameters, and self-discharge characterization parameters.

[0035] In an exemplary embodiment of this application, step S300, determining the target parameter set for each individual cell based on the charging evaluation data source and the discharging evaluation data source, further includes: obtaining the internal resistance parameters and open-circuit voltage parameters of each individual cell in the battery pack through battery model parameter identification based on the charging evaluation data source; obtaining the self-discharge characterization parameters of each individual cell based on the discharging evaluation data source; and obtaining the voltage parameters and temperature parameters of each individual cell based on the charging evaluation data source and the discharging evaluation data source, respectively.

[0036] In an exemplary embodiment of this application, the step of identifying the internal resistance parameters and open-circuit voltage parameters of each individual cell in the battery pack based on the charging evaluation data source and battery model parameters further includes: firstly, constructing a first-order RC equivalent circuit model using the terminal voltage, current, and time data from the charging evaluation data source, and treating the open-circuit voltage as a constant value within a short time window where the change in the state of charge is less than a preset threshold; then, deriving the terminal voltage difference equation based on the first-order RC equivalent circuit model, and using a recursive least squares algorithm to identify the internal resistance parameters and open-circuit voltage parameters of the battery pack online; finally, using the voltage divider principle of a series circuit, calculating the internal resistance parameters and open-circuit voltage parameters of each individual cell in the battery pack based on the identified internal resistance parameters and open-circuit voltage parameters of the battery pack.

[0037] It should be noted that, based on the similar operating condition segments obtained in step S100, this application evaluates the consistency of five core battery parameters, including voltage parameters, temperature parameters, internal resistance parameters, open-circuit voltage parameters, and self-discharge characterization parameters (K-value). The voltage and temperature parameters of a single cell can be directly obtained from the raw data uploaded by the sensors, while the internal resistance and open-circuit voltage parameters at the single cell level need to be obtained indirectly through modeling and identification.

[0038] Specifically, please refer to Figure 2 As shown, a first-order RC equivalent circuit model is constructed using the terminal voltage, current, and time data from the charging evaluation data source. Within a short time window where the SOC change is less than a preset threshold, the open-circuit voltage is treated as a constant value to reduce the impact of SOC estimation bias and battery aging drift on parameter identification accuracy.

[0039] Based on the dynamic characteristics of the first-order RC equivalent circuit model, its continuous domain state equation and output equation can be expressed as follows:

[0040]

[0041] in, Indicates open-circuit voltage. Indicates the internal resistance of the ohm. and These represent the polarization resistance and polarization capacitance, respectively. This represents the voltage across the polarized capacitor. Indicates terminal voltage. It represents electric current.

[0042] Discretize the above two equations within a selected short time window, and assume an open-circuit voltage. Under the premise of keeping the voltage constant, the differential equation of the terminal voltage at adjacent sampling times can be derived as follows:

[0043] The above terminal voltage difference equation can be rearranged into a linear regression form:

[0044] in:

[0045]

[0046]

[0047] In the formula, This represents the voltage difference between adjacent time steps. For the input vector, The parameter vector to be identified.

[0048] To achieve online parameter identification, this embodiment employs a recursive least squares algorithm to analyze the parameter vector. Dynamic estimation is performed. The recursive least squares algorithm can update the parameter estimates in real time as new data is collected, without the need to store a complete historical dataset, thus meeting the requirements of computational efficiency and real-time performance in practical application scenarios.

[0049] It should be noted that in this embodiment, the battery pack is composed of multiple individual battery cells connected in series. The equivalent ohmic internal resistance at the battery pack level is obtained through the above identification process. With equivalent open circuit voltage Then, based on the voltage divider principle of series circuits, the ohmic internal resistance and open-circuit voltage of each individual cell are estimated using the following formulas:

[0050]

[0051] in, This indicates the total terminal voltage of the battery pack. Indicates the first at the same time The measured terminal voltage values ​​of individual battery cells are used. Using the above method, the internal resistance and open-circuit voltage parameters of each individual cell can be deduced from the parameter identification results at the battery pack level, providing a data basis for subsequent consistency evaluation of individual cells.

[0052] It should be noted that the self-discharge characterization parameter is the open-circuit voltage decay rate. In an exemplary embodiment of this application, obtaining the self-discharge characterization parameter of each individual cell according to the discharge evaluation data source includes: for each individual cell, selecting from the discharge evaluation data source multiple discharge segments whose starting points satisfy zero current and a state of charge greater than a preset state of charge threshold; determining discharge segment pairs with a time interval greater than or equal to a preset time threshold from the selected multiple discharge segments, and for each discharge segment pair, using the voltage at the starting points of the two discharge segments as the open-circuit voltage, calculating the single open-circuit voltage decay rate based on the ratio of the difference between the two open-circuit voltages to the time interval; and determining the open-circuit voltage decay rate of the individual cell based on the statistical values ​​of the multiple single open-circuit voltage decay rates corresponding to the individual cell.

[0053] Specifically, to quantify the self-discharge level of each individual cell, this application introduces the open-circuit voltage decay rate index (K value) as a core evaluation parameter. The K value is used to characterize the rate at which the open-circuit voltage of an individual cell decreases over time in a static state. Its calculation is based on the open-circuit voltage values ​​corresponding to two independent static moments that meet specific constraints.

[0054] The formula for calculating the K value is defined as follows:

[0055] in, and They represent the first Save individual battery cells at all times and The effective open-circuit voltage value, in volts (V); Indicates time and The time interval between the two open-circuit voltages is expressed in hours (h). In the formula, the numerator is the absolute value of the difference between the two open-circuit voltages; multiplying by 1000 converts the voltage unit from volts to millivolts; dividing by the time interval... After multiplying by 24 hours, the final K value is in millivolts per day (mV / day), which is used to characterize the average daily voltage decay of the single cell under static conditions.

[0056] It should be noted that the open-circuit voltage value and The selection must simultaneously meet the following constraints: First, the two open-circuit voltage values ​​must correspond to the start times of two discharge segments under similar operating conditions obtained through step S200; Second, the time interval between the two values ​​must be... The K-value is determined by several factors. First, the minimum resting time threshold must be greater than or equal to a preset threshold to ensure that the self-discharge effect produces a sufficiently identifiable cumulative difference in voltage variation. Second, at the start of the discharge segment, the measured current value of the battery pack must be zero amperes (0A) to eliminate dynamic deviations in terminal voltage caused by ohmic internal resistance and polarization effects, thereby ensuring that the terminal voltage at that moment can be approximated as the true open-circuit voltage. Third, at the start of the discharge segment, the SOC of the battery pack must be higher than a preset state-of-charge threshold to reduce the impact of differences in the slope of the open-circuit voltage-state-of-charge curve within different SOC ranges on the consistency of K-value calculation, making the K-value calculation result more accurately reflect the self-discharge behavior of individual cells. It is worth noting that in this embodiment, the preset state-of-charge threshold is 95%. Through the above calculation formula and strict data screening conditions, the obtained K-value can accurately characterize the relative self-discharge rate of each individual cell. Under normal aging conditions, the K-value of a high-performance cell typically does not exceed 2mV / day; if the K-value of a certain individual cell exceeds this range, it indicates that it has an abnormal self-discharge risk.

[0057] Next, step S400 is executed, which involves extracting and aggregating consistent statistical features of the target parameter set to generate a multidimensional evaluation vector.

[0058] In an exemplary embodiment of this application, step S400, which involves extracting and aggregating consistent statistical features of the target parameter set to generate a multidimensional evaluation vector, further includes: extracting consistent statistical features from the voltage parameter, the temperature parameter, the internal resistance parameter, the open-circuit voltage parameter, and the self-discharge characterization parameter, and aggregating the consistent statistical features to generate a multidimensional evaluation vector; the consistent statistical features include standard deviation, range, coefficient of variation, mean entropy, and maximum entropy.

[0059] It should be noted that, in this embodiment, to effectively track the evolution of consistency among individual cells within the battery pack over time, this step, based on the data obtained in steps S200 and S300, performs monthly-level consistency statistical feature extraction and hierarchical aggregation processing on five core evaluation parameters. The consistency statistical features include standard deviation, range, coefficient of variation, mean entropy, and maximum entropy. The five core evaluation parameters, according to their respective data characteristics and evaluation requirements, originate from both the charging evaluation data source and the discharging evaluation data source. The consistency evaluation of the voltage and temperature parameters utilizes both the charging and discharging evaluation data sources. Specifically, the charging segment evaluation data is based on similar charging segments obtained through clustering, and the discharging segment evaluation data is based on similar discharging segments obtained through screening. The evaluation results of the charging and discharging segments are jointly aggregated at the monthly level. The consistency evaluation of the internal resistance and open-circuit voltage parameters is based solely on the parameter data identified from the charging evaluation data source; the consistency evaluation of the self-discharge characterization parameter (i.e., the K value) is based solely on the parameter data calculated from the discharging evaluation data source.

[0060] For the four types of sampling point-level parameters—voltage, temperature, internal resistance, and open-circuit voltage—that is, the current parameter value of each individual cell at each sampling moment, the monthly consistency features are extracted as follows: At each sampling moment within each similar operating condition charging segment, based on the current parameter values ​​of all individual cells at that moment, five consistency statistical features (standard deviation, range, coefficient of variation, mean entropy, and maximum entropy) are calculated to characterize the consistency level among individual cells at that sampling moment. For each feature value obtained at all sampling moments within the same charging segment, aggregation is performed to generate the five consistency feature values ​​corresponding to that charging segment. The aggregation rules are: for the standard deviation, range, coefficient of variation, and mean entropy features, the arithmetic mean of the feature values ​​corresponding to all sampling moments within the charging segment is taken; for the maximum entropy feature, the maximum value among the feature values ​​corresponding to all sampling moments is taken. At the monthly aggregation level, only the cluster with the largest sample size after operating condition clustering in step S200 (i.e., the most representative charging operating condition within the month) containing all charging segments is selected as the aggregation basis. The five consistency feature values ​​generated from all eligible charging segments within the month are then aggregated again. Finally, the voltage parameter, temperature parameter, internal resistance parameter, and open-circuit voltage parameter each form a monthly consistency vector composed of five consistency statistical features.

[0061] For the self-discharge characterization parameter, in this embodiment, the open-circuit voltage decay rate (K value) is a segment-level parameter because its calculation depends on the open-circuit voltage decay between two effective resting moments. Its consistency assessment is performed directly on a segment-by-segment basis. Specifically, for each eligible discharge segment, its K value is first calculated based on each individual cell. Then, five consistency statistical characteristics (standard deviation, range, coefficient of variation, mean entropy, and maximum entropy) are directly calculated within that discharge segment as its self-discharge consistency characterization. During monthly aggregation, the five characteristic values ​​obtained from all eligible discharge segments within the month are aggregated to obtain the five consistency characteristic values ​​of the self-discharge characterization parameter for that month. Through the above hierarchical calculation and aggregation process, the battery state for each month is ultimately represented as a 25-dimensional monthly consistency feature vector. This monthly consistency feature vector comprehensively reflects the consistency of each individual cell in the battery pack across multiple dimensions, including electrical characteristics, thermal characteristics, and aging characteristics, providing a structured input data foundation for subsequent comprehensive evaluation and quantitative scoring.

[0062] Finally, step S500 is executed, which involves determining the weights of each feature dimension in the multidimensional evaluation vector and performing a comprehensive evaluation of the multidimensional evaluation vector based on the determined weights to output a quantitative evaluation result of battery consistency.

[0063] In an exemplary embodiment of this application, step S500, determining the weights of each feature dimension in the multidimensional evaluation vector, further includes: constructing a fuzzy evaluation matrix for each feature dimension with respect to a preset consistency evaluation level; calculating the information entropy and difference coefficient of each feature dimension based on the fuzzy evaluation matrix, and determining the objective weights of each feature dimension based on the information entropy and the difference coefficients; sorting each feature dimension in descending order of objective weight, selecting the top few feature dimensions whose cumulative weight contribution rate reaches a preset proportion as core indicators, and normalizing the weights of the selected core indicators to obtain the final weights of each feature dimension used for comprehensive evaluation. It should be noted that the preset consistency evaluation level is, in order, high consistency, relatively high consistency, moderate consistency, and low consistency.

[0064] Specifically, in this embodiment, the entropy weight method is used to determine the weight of each feature dimension, and the fuzzy comprehensive evaluation method is combined to quantify and score the monthly consistency status of the battery.

[0065] For each consistency characteristic under each parameter category, the membership relationship between the characteristic value and the preset consistency evaluation level is first established. This application divides the battery consistency status into four evaluation levels: high consistency, relatively high consistency, average consistency, and low consistency.

[0066] For the Features ( ,in By using a pre-selected membership function, the observed values ​​of this feature are mapped to the four evaluation levels mentioned above, resulting in the corresponding membership vector:

[0067] in, Indicates the first The feature for the first The degree of membership of each evaluation level, and satisfying the normalization condition:

[0068] All The membership vectors of each feature are combined row by row to form the fuzzy evaluation matrix. :

[0069] in, For the total number of features, The evaluation level is determined by the number of grades.

[0070] After obtaining the fuzzy evaluation matrix Based on this, the entropy weight method is used to calculate the objective weights of each feature dimension to reflect the differences in the amount of information carried by different features in distinguishing battery consistency states. The specific calculation steps are as follows: First, calculate the... Information entropy of each feature :

[0071] In the formula, For the evaluation rating number. When At that time, it was agreed Information entropy The range of values ​​is The smaller the entropy value, the greater the difference in membership degree of the feature across different evaluation levels.

[0072] Next, calculate the first... Difference coefficient of each feature The difference coefficient characterizes the effectiveness of the information contained in the feature, and is defined as follows:

[0073] Coefficient of difference The larger the value, the greater the potential contribution of this feature to the overall evaluation.

[0074] Finally, the difference coefficients are normalized, and the first... Initial objective weights of each feature :

[0075] in, Indicates the first The difference coefficients of each feature, and the resulting weights satisfy the following: .

[0076] After calculating the initial objective weights for all 25 feature dimensions using the entropy weight method described above, this embodiment introduces a core indicator screening mechanism to simplify the evaluation model and enhance the interpretability of the evaluation results. Given that the weights of some feature dimensions may be too low, their marginal contribution to the final evaluation result is extremely limited, and they can be removed.

[0077] Specifically, the weights of all 25 features are... Sort in descending order from largest to smallest, and calculate the first... Cumulative weight contribution rate of each feature dimension ,in, Indicates the sorted order of the first... The weights of each feature dimension are assigned. A cumulative weight contribution rate threshold is set (e.g., 85% or 90%), and the top few feature dimensions whose cumulative contribution rate reaches this threshold are retained as core evaluation indicators. For the selected subset of core indicators, their original objective weights are re-normalized to obtain the indicator weight vector used for the final comprehensive evaluation. To ensure that the sum of its components is 1, the weights of the indicators belonging to the same parameter category (i.e., voltage parameter, temperature parameter, internal resistance parameter, open-circuit voltage parameter, and K value) in the core indicator subset are summed to obtain the total weight allocation of the five parameter categories in the comprehensive evaluation. Based on the membership vector of the retained core indicators, combined with the renormalized weight vector, the fuzzy comprehensive evaluation operator is used to calculate the comprehensive membership degree of each vehicle's monthly battery consistency at the four evaluation levels. The comprehensive membership degree is then further transformed into a quantitative score, which is output as the final battery consistency evaluation result.

[0078] In summary, this invention provides a battery consistency evaluation method and system based on clustering and screening under similar operating conditions. The method includes: acquiring actual operating data and segmenting it into multiple charging and discharging segments according to charging and discharging states; extracting current curve features of the charging segments and performing operating condition clustering to obtain charging evaluation data sources; screening discharging segments based on preset state thresholds to obtain discharging evaluation data sources; identifying the internal resistance and open-circuit voltage parameters of each individual cell based on the charging evaluation data sources, obtaining self-discharge characterization parameters based on the discharging evaluation data sources, and obtaining voltage and temperature parameters based on both types of data sources; extracting consistency statistical features from the five types of parameters (internal resistance, open-circuit voltage, self-discharge, voltage, and temperature) of the acquired individual cells, aggregating them monthly, and generating a multi-dimensional evaluation vector; determining the weight of each feature dimension and comprehensively evaluating the multi-dimensional evaluation vector to output a quantitative evaluation result of battery consistency. This invention extracts charging segments under similar operating conditions from actual operating data through cluster analysis and screens discharging segments under a baseline state, effectively removing the coupling interference of operating condition factors such as charge / discharge rate, SOC range, and temperature on consistency indicators, ensuring that the evaluation results truly reflect the differences in the cell's intrinsic parameters. Based on differentiated processing of charging and discharging segments, online identification and extraction of indirect parameters such as internal resistance, open-circuit voltage, and self-discharge decay rate are achieved. This overcomes the limitations of existing methods that can only evaluate directly measurable parameters such as voltage and temperature, significantly expanding the dimensions and depth of consistency assessment. The entropy weighting method is used to objectively assign weights to each feature dimension and aggregate them to generate a multi-dimensional evaluation vector, achieving a comprehensive quantitative evaluation of battery consistency. The evaluation results are more robust and have greater engineering practical value.

[0079] Based on the same inventive concept, please refer to Figure 3 As shown, another embodiment of the present invention also provides a battery consistency evaluation system 100 based on clustering and screening of similar operating conditions, implemented using the battery consistency evaluation method based on clustering and screening of similar operating conditions as described in any of the above embodiments, including: The battery data acquisition module 110 is used to acquire the actual operating data of the battery and divide the actual operating data into multiple charging segments and multiple discharging segments according to the charging and discharging state. The similar operating condition extraction module 120 is used to perform operating condition clustering processing on multiple charging segments to obtain charging evaluation data sources under similar operating conditions; and to perform state filtering processing on multiple discharging segments to obtain discharging evaluation data sources under similar operating conditions. The parameter identification module 130 is used to determine the target parameter set of each individual cell based on the charging evaluation data source and the discharging evaluation data source. The target parameter set includes voltage parameters, temperature parameters, internal resistance parameters, open circuit voltage parameters, and self-discharge characterization parameters. The feature aggregation module 140 is used to extract and aggregate consistent statistical features from the target parameter set to generate a multidimensional evaluation vector. The quantitative evaluation module 150 is used to determine the weight of each feature dimension in the multidimensional evaluation vector, and to comprehensively evaluate the multidimensional evaluation vector according to the determined weights, so as to output the quantitative evaluation result of battery consistency.

[0080] It should be noted that, in this embodiment, the battery consistency evaluation system 100 based on similar operating conditions clustering and screening includes the battery consistency evaluation method based on similar operating conditions clustering and screening as described in any of the above embodiments. Since the battery consistency evaluation system 100 based on similar operating conditions clustering and screening provided in this embodiment belongs to the same inventive concept as the battery consistency evaluation method provided in any of the above embodiments, it has at least the same beneficial effects, and will not be described in detail here.

[0081] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A battery consistency evaluation method based on similar working condition clustering and screening, characterized in that, include: Acquire the actual operating data of the battery, and divide the actual operating data into multiple charging segments and multiple discharging segments according to the charging and discharging state; Multiple charging segments are subjected to working condition clustering to obtain charging evaluation data sources under similar working conditions; Furthermore, a state filtering process is performed on multiple discharge segments to obtain discharge assessment data sources under similar operating conditions; Based on the charging evaluation data source and the discharging evaluation data source, the target parameter set for each individual cell is determined. The target parameter set includes voltage parameters, temperature parameters, internal resistance parameters, open circuit voltage parameters, and self-discharge characterization parameters. Consistency statistical features are extracted and aggregated from the target parameter set to generate a multidimensional evaluation vector; The weights of each feature dimension in the multidimensional evaluation vector are determined, and the multidimensional evaluation vector is comprehensively evaluated based on the determined weights to output a quantitative evaluation result of battery consistency.

2. The battery consistency evaluation method based on clustering and screening under similar operating conditions according to claim 1, characterized in that, The step of performing operating condition clustering processing on multiple charging segments to obtain charging evaluation data sources under similar operating conditions includes: Extract the current curve features of each charging segment, including current intensity features, current stability features, current distribution features, and average rate of current change features. Based on the characteristics of the current curve, multiple charging segments are clustered according to their operating conditions to obtain charging segments under similar operating conditions as the charging evaluation data source.

3. The battery consistency evaluation method based on clustering and screening under similar operating conditions according to claim 2, characterized in that, The step of performing operating condition clustering processing on multiple charging segments based on the current curve characteristics to obtain charging segments under similar operating conditions as the charging evaluation data source includes: Based on the current curve characteristics of each charging segment, a first feature vector corresponding to each charging segment is constructed; The first feature vector corresponding to each charging segment is standardized to obtain the second feature vector corresponding to each charging segment. A density-based noise-based spatial clustering algorithm is used to perform unsupervised clustering on the second feature vectors corresponding to each charging segment to obtain at least one cluster. The cluster with the largest number of samples is selected as the similar working condition cluster, and each charging segment corresponding to the similar working condition cluster is used as the charging evaluation data source.

4. The battery consistency evaluation method based on clustering and screening under similar operating conditions according to claim 1, characterized in that, The step of performing state filtering on multiple discharge segments to obtain discharge assessment data sources under similar operating conditions includes: Discharge segments whose state of charge variation range covers a preset state of charge range and whose individual cell temperatures are within a preset temperature range are selected from multiple discharge segments as the discharge evaluation data source under similar operating conditions.

5. The battery consistency evaluation method based on clustering and screening under similar operating conditions according to claim 1, characterized in that, The step of determining the target parameter set for each individual battery cell based on the charging evaluation data source and the discharging evaluation data source includes: Based on the charging evaluation data source, the internal resistance parameters and open-circuit voltage parameters of each individual cell in the battery pack are obtained by identifying the battery model parameters. The self-discharge characterization parameters of each individual cell are obtained based on the discharge evaluation data source. The voltage and temperature parameters of each individual battery cell are obtained based on the charging evaluation data source and the discharging evaluation data source, respectively.

6. The battery consistency evaluation method based on clustering and screening under similar operating conditions according to claim 5, characterized in that, The step of identifying and obtaining the internal resistance parameters and open-circuit voltage parameters of each individual cell in the battery pack based on the charging evaluation data source and through battery model parameter identification includes: A first-order RC equivalent circuit model is constructed using the terminal voltage, current, and time data from the charging evaluation data source, and the open-circuit voltage is regarded as a constant value within a short time window when the change in the state of charge is less than a preset threshold. The terminal voltage difference equation is derived based on the first-order RC equivalent circuit model, and the internal resistance parameters and open-circuit voltage parameters of the battery pack are identified online using the recursive least squares algorithm. By utilizing the voltage divider principle of series circuits, and based on the identified internal resistance parameters and open-circuit voltage parameters of the battery pack, the internal resistance parameters and open-circuit voltage parameters of each individual cell in the battery pack are calculated.

7. The battery consistency evaluation method based on clustering and screening under similar operating conditions according to claim 5, characterized in that, The self-discharge characterization parameter is the open-circuit voltage decay rate; The step of obtaining the self-discharge characterization parameters of each individual cell based on the discharge evaluation data source includes: For each individual battery cell, multiple discharge segments are selected from the discharge evaluation data source, where the starting point corresponding to the individual battery cell satisfies zero current and a state of charge greater than a preset state of charge threshold. From the selected multiple discharge segments, determine the discharge segment pairs with a time interval greater than or equal to a preset time threshold, and for each discharge segment pair, use the voltage at the starting point of the two discharge segments as the open circuit voltage, and calculate the single open circuit voltage decay rate based on the ratio of the difference between the two open circuit voltages to the time interval. The open-circuit voltage decay rate of the individual cell is determined based on the statistical values ​​of the multiple single open-circuit voltage decay rates corresponding to the individual cell.

8. The battery consistency evaluation method based on clustering and screening under similar operating conditions according to claim 1, characterized in that, The step of extracting and aggregating consistent statistical features from the target parameter set to generate a multidimensional evaluation vector includes: Consistency statistical features are extracted from the voltage parameter, temperature parameter, internal resistance parameter, open-circuit voltage parameter, and self-discharge characterization parameter, respectively, and the consistency statistical features are aggregated to generate a multidimensional evaluation vector; the consistency statistical features include standard deviation, range, coefficient of variation, mean entropy, and maximum entropy.

9. The battery consistency evaluation method based on clustering and screening under similar operating conditions according to claim 1, characterized in that, Determining the weights of each feature dimension in the multidimensional evaluation vector includes: Construct a fuzzy evaluation matrix for each feature dimension with respect to the preset consistency evaluation level; Based on the fuzzy evaluation matrix, calculate the information entropy and difference coefficient of each feature dimension, and determine the objective weight of each feature dimension based on the information entropy and the difference coefficient. The feature dimensions are sorted from largest to smallest according to their objective weights. The top few feature dimensions whose cumulative weight contribution rate reaches a preset ratio are selected as core indicators. The weights of the selected core indicators are then normalized to obtain the final weights of each feature dimension used for comprehensive evaluation.

10. A battery consistency evaluation system based on clustering and screening under similar operating conditions, characterized in that, The battery consistency evaluation method based on clustering and screening under similar operating conditions, as described in any one of claims 1 to 9, is adopted, including: The battery data acquisition module is used to acquire the actual operating data of the battery and divide the actual operating data into multiple charging segments and multiple discharging segments according to the charging and discharging state. The similar operating condition extraction module is used to perform operating condition clustering processing on multiple charging segments to obtain charging evaluation data sources under similar operating conditions; and to perform state filtering processing on multiple discharging segments to obtain discharging evaluation data sources under similar operating conditions. The parameter identification module is used to determine the target parameter set of each individual cell based on the charging evaluation data source and the discharging evaluation data source. The target parameter set includes voltage parameters, temperature parameters, internal resistance parameters, open circuit voltage parameters, and self-discharge characterization parameters. The feature aggregation module is used to extract and aggregate consistent statistical features from the target parameter set to generate a multidimensional evaluation vector. The quantitative evaluation module is used to determine the weight of each feature dimension in the multidimensional evaluation vector, and to comprehensively evaluate the multidimensional evaluation vector according to the determined weights, so as to output the quantitative evaluation result of battery consistency.

Images