Energy storage battery capacity analysis method and device, electronic equipment and storage medium

Abstract

The application provides a method and device for analyzing the capacity of an energy storage battery, electronic equipment and a storage medium. The method comprises: obtaining actual operation data of an energy storage battery system at a current time; obtaining a current calibration point according to the actual operation data and a previous analysis model; adding the current calibration point and a corresponding calibration point weight to a previous curve fitting data set, and updating the historical weights of each historical calibration point in the previous curve fitting data set according to a preset weight decay factor to obtain a current curve fitting data set; fitting each calibration point in the current curve fitting data set to obtain a current analysis model; and analyzing the battery capacity of the energy storage battery system based on the current analysis model if the current analysis model meets a preset convergence condition. The application uses the charging and discharging data of the actual operation of the energy storage system to establish an accurate SOC-OCV function relationship, is adaptive to different battery systems and aging degrees, has high calibration accuracy, and has small calculation amount.

Description

Technical Field

[0001] This disclosure relates to the field of battery technology, and in particular to a method, apparatus, electronic device and storage medium for analyzing the capacity of energy storage batteries. Background Technology

[0002] As the scale of renewable energy power generation continues to expand, electrochemical energy storage systems are playing an increasingly important role in grid peak shaving, frequency regulation, and power output smoothing. Lithium-ion batteries, due to their high energy density and long cycle life, have become the mainstream technology for large-scale energy storage.

[0003] Battery capacity is a key indicator for assessing battery state of health (SOH) and the energy management capabilities of energy storage systems. Accurate estimation of individual cell capacity is not only crucial for the scheduling strategy and safe operation of energy storage systems, but also fundamental for identifying lagging cells, guiding equalization maintenance, and predicting remaining lifespan. However, due to differences in manufacturing processes, cycle aging, temperature fluctuations, and other factors, the actual capacity of different cells within the same battery cluster varies significantly. Traditional assumptions of uniform rated capacity can lead to capacity estimation errors, thereby affecting overall system performance and safety decisions.

[0004] Traditional capacity estimation methods have the following problems: (1) Requires complete charge-discharge cycle testing: Traditional methods require the battery to undergo a complete constant current charge-discharge cycle in the laboratory to obtain accurate capacity through ampere-hour integration. In actual operation, energy storage systems often operate in a partial charge-discharge range, which cannot provide a complete SOC window. Furthermore, capacity decay after aging causes test benchmark drift, increasing estimation error.

[0005] (2) High cost of offline calibration: The battery needs to be removed from the energy storage system and sent to the laboratory for offline capacity testing, which is time-consuming and labor-intensive, affects the normal operation of the system, and makes it impossible to achieve online capacity monitoring during the operation of the energy storage power station.

[0006] (3) Inability to adapt to individual cell differences: Due to different degrees of attenuation, the actual capacity distribution of cells within the same cluster tends to be discrete. Using a uniform rated capacity as a benchmark cannot reflect the true capacity differences of individual cells, making it difficult to identify lagging cells and resulting in large capacity calculation errors.

[0007] (4) Lack of online capacity estimation methods: Traditional methods rely on the fixed rated capacity reported by BMS or the results of periodic offline sampling inspections. They cannot use the charging and discharging data generated in daily operation to dynamically update the capacity estimation. The SOC calibration error will be further transmitted to the capacity calculation, causing the capacity estimation accuracy to degrade over time.

[0008] To address these issues, researchers have proposed various data-driven capacity estimation methods in recent years, including model-based prediction, machine learning regression, and ampere-hour integral correction. However, most of these methods require large amounts of labeled training data or rely on complex neural network models, resulting in high computational complexity, poor interpretability, and difficulty in deployment and validation in practical BMS systems. Summary of the Invention

[0009] This disclosure provides a method, apparatus, electronic device, and storage medium for analyzing the capacity of energy storage batteries. It utilizes the charge and discharge data from the actual operation of the energy storage system to establish an accurate SOC-OCV function relationship, which is adaptive to different battery systems and aging levels, with high calibration accuracy and low computational load.

[0010] In a first aspect, a method for analyzing the capacity of an energy storage battery, the method comprising: Obtain the current operating data of the energy storage battery system; The current calibration point is obtained based on the actual operating data and the previous analysis model. The current calibration point is a SOC-OCV data point. The previous analysis model includes the previous curve fitting dataset. The curve fitting dataset includes several historical calibration points and the historical weights of each historical calibration point. The current calibration point and its corresponding calibration point weights are added to the previous curve fitting dataset, and the historical weights of each historical calibration point in the previous curve fitting dataset are updated according to the preset weight decay factor to obtain the current curve fitting dataset. The current analysis model is obtained by fitting each calibration point in the current curve fitting dataset, and the previous analysis model is obtained by fitting each calibration point in the previous curve fitting dataset. If the current analysis model meets the preset convergence conditions, then the battery capacity of the energy storage battery system is analyzed based on the current analysis model.

[0011] The above energy storage battery capacity analysis method uses the actual charging and discharging data of the energy storage system to establish an accurate SOC-OCV function relationship, which is adaptive to different battery systems and aging levels, with high calibration accuracy and low computational load.

[0012] Optionally, before obtaining the actual operating data of the energy storage battery system at the current moment, the following steps are also included: Acquire the initial operating data of the energy storage battery system at the initial moment, where the initial moment is the first moment when the generative analysis model is started; The initial running data is aligned to a reference, and the initial calibration point is obtained by combining it with a preset prior curve. The initial analysis model is obtained by fitting the initial calibration points.

[0013] Optionally, the step of aligning the initial running data to a reference and combining it with a preset prior curve to obtain initial calibration points includes: The initial operating data is aligned to a reference to determine the battery information at the discharge cutoff time and the charging cutoff time. The battery information includes open-circuit voltage information and charge information. Prior data points are obtained by integrating the initial running data and the preset prior curve; The initial calibration point is determined based on the battery information at the discharge cutoff time and the charging cutoff time, and the prior data points.

[0014] Optionally, the step of performing reference alignment on the initial operating data, wherein the initial operating data determines battery information at the discharge cutoff time and the charge cutoff time, includes: Based on the initial operating data, determine the voltage data, charge data, and total cluster current data of each cell; The charge / discharge cycle is identified based on the total cluster current data; The voltage data of each cell, the charge data, and the charge / discharge cycle are aligned to a reference to determine the battery information at the discharge cutoff time and the charging cutoff time.

[0015] Optionally, before determining the battery information at the discharge cutoff time and charge cutoff time by performing reference alignment on the voltage data of each cell, the charge data, and the charge / discharge cycle, the step further includes: Identify at least two complete charge-discharge cycles, and determine the open-circuit voltage of each cell at the charge-discharge cutoff time based on the voltage data of each cell; The calibration reference period is determined from the at least two complete charge-discharge cycles based on the open-circuit voltage of each cell at the charge-discharge cutoff time. The step involves aligning the voltage data of each cell, the charge data, and the charge / discharge cycle to a reference, and determining the battery information at the discharge cutoff time and the charge cutoff time, including: The voltage data of each cell, the charge data, and the charge / discharge cycle within the calibration reference period are aligned to determine the battery information at the discharge cutoff time and the charge cutoff time.

[0016] Optionally, the preset convergence condition is set to the average capacity change being less than a preset threshold.

[0017] Optionally, if the current analysis model meets the preset convergence condition, then the battery capacity of the energy storage battery system is analyzed based on the current analysis model, including: If the current analysis model meets the preset convergence conditions, then the cumulative charge during the discharge process, as well as the charge change at the start and end of the discharge, are determined based on the current analysis model. The battery capacity is determined based on the cumulative charge and the change in charge.

[0018] Secondly, a battery capacity analysis device is provided, the device comprising: The data acquisition module is used to acquire the actual operating data of the energy storage battery system at the current moment; The data analysis module is used to obtain the current calibration point based on the actual operating data and the previous analysis model. The current calibration point is a SOC-OCV data point. The previous analysis model includes the previous curve fitting dataset. The curve fitting dataset includes several historical calibration points and the historical weights of each historical calibration point. The model analysis module is used to add the current calibration point and its corresponding calibration point weights to the previous curve fitting dataset, and update the historical weights of each historical calibration point in the previous curve fitting dataset according to a preset weight decay factor to obtain the current curve fitting dataset; and to fit each calibration point in the current curve fitting dataset to obtain the current analysis model, wherein the previous analysis model is obtained by fitting each calibration point in the previous curve fitting dataset. The capacity analysis module is used to analyze the battery capacity of the energy storage battery system based on the current analysis model if the current analysis model meets the preset convergence conditions.

[0019] Thirdly, an electronic device is provided, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the energy storage battery capacity analysis method provided in the first aspect above.

[0020] Fourthly, a computer-readable storage medium is provided, characterized in that the computer-readable storage medium includes a computer program, which, when executed, controls the electronic device in which the computer-readable storage medium is located to perform the energy storage battery capacity analysis method provided in the first aspect above. Attached Figure Description

[0021] The accompanying drawings used in the description of the embodiments of this disclosure are briefly introduced below: Figure 1 A schematic flowchart of a battery capacity analysis method provided in some embodiments of this application is shown; Figure 2 This document illustrates a schematic diagram of weighted B-spline fitting provided in some embodiments of this application. Figure 3 A schematic diagram of another energy storage battery capacity analysis device provided in some embodiments of this application is shown. Detailed Implementation

[0022] To more clearly illustrate the technical solutions in the embodiments of this disclosure, the specific implementation methods of this disclosure will be described below with reference to the accompanying drawings. The accompanying drawings described below are merely some embodiments of this disclosure. For those skilled in the art, other drawings or embodiments can be obtained based on these drawings or embodiments without creative effort. Adjustments and improvements made without departing from the concept of this disclosure are all within the protection scope of this disclosure.

[0023] To keep the drawings simple, each figure only schematically shows the parts relevant to the embodiment, and they do not represent the actual structure of the product. In addition, for the sake of simplicity and ease of understanding, only some structures or parts are schematically shown, and there may be more or fewer similar structures or parts in reality.

[0024] Figure 1 A schematic flowchart of a battery capacity analysis method provided in some embodiments of this application is shown. The battery capacity analysis method includes at least the following steps: S110: Obtain the actual operating data of the energy storage battery system at the current moment; S120: The current calibration point is obtained based on the actual operating data and the previous analysis model. The current calibration point is a SOC-OCV data point. The previous analysis model includes the previous curve fitting dataset. The curve fitting dataset includes several historical calibration points and the historical weights of each historical calibration point. S130: Add the current calibration point and its corresponding calibration point weight to the previous curve fitting dataset, and update the historical weight of each historical calibration point in the previous curve fitting dataset according to the preset weight decay factor to obtain the current curve fitting dataset. S140: Fit each calibration point in the current curve fitting dataset to obtain the current analysis model, wherein the previous analysis model is obtained by fitting each calibration point in the previous curve fitting dataset; S150: If the current analysis model meets the preset convergence conditions, then analyze the battery capacity of the energy storage battery system based on the current analysis model.

[0025] Specifically, the actual operating data of the energy storage battery system at the current moment is obtained. The charge / discharge cycle is identified based on the actual operating data, and the open-circuit voltage (OCV) of each cell at the end of the charge / discharge cycle is extracted. The actual operating data consists of data related to each cell in the energy storage battery system, including but not limited to cell voltage data and total cluster current data. The start and end events of charge / discharge are identified based on changes in current direction: start of discharge: current changes from less than or equal to Ith to greater than Ith; end of discharge: current changes from greater than Ith to less than or equal to Ith; start of charging: current changes from greater than or equal to negative Ith to less than negative Ith; end of charging: current changes from less than negative Ith to greater than or equal to negative Ith; where Ith is a preset current threshold, ranging from 1A to 20A. Short cycles with a duration less than Tmin are filtered out; Tmin ranges from 1 to 60 minutes, and the two longest charge / discharge cycles are selected for calibration. After the charge / discharge cycle ends, the average value of the last N voltage data points within the resting period is selected as the OCV, where N is an integer from 3 to 10, and the resting time window is 30 to 300 minutes. It should be noted that the specific values ​​of various thresholds in this embodiment are distance descriptions, and can be freely set according to actual needs.

[0026] The analysis model is the SOC-OCV model. The previous analysis model was obtained by fitting the curve fitting dataset based on the calibration points in the previous curve fitting dataset. The curve fitting dataset includes several historical calibration points and their historical weights. For example, the curve fitting dataset can be set as a weighted (OCV, SOC, weight) triplet set, where the weight represents the confidence level of the corresponding calibration point in the model fitting process. Inputting relevant data from the actual running data into the previous analysis model yields the current calibration point, which is the SOC-OCV data point. Each generation of a new analysis model constitutes an iteration, and the generation steps are identical.

[0027] The current calibration point and its corresponding calibration point weights are added to the previous curve fitting dataset. Simultaneously, the historical weights of each historical calibration point in the previous curve fitting dataset are updated according to a preset weight decay factor, resulting in the current curve fitting dataset. The weight decay factor alpha is used to configure the decay rate of the historical calibration point weights during iteration, and its value ranges from 0.90 to 0.99.

[0028] In each iteration, the newly generated current calibration point is added to the curve fitting dataset with a full weight of 1.0, meaning the weight of the current calibration point can be set to 1.0. The weights of historical calibration points decay exponentially: weight_new = weight_old x alpha; the default alpha = 0.99, corresponding to a weight decay to 50% of its original value after approximately 69 iterations. When the alpha value is smaller (e.g., 0.90), historical calibration points are forgotten more quickly, suitable for scenarios with rapidly changing data distributions; when the alpha value is larger (e.g., 0.99), historical calibration points are retained longer, suitable for scenarios with stable data distributions. Through the old point weight decay mechanism, the algorithm can automatically reduce its trust in potentially inaccurate early calibration points during iteration, gradually increasing its reliance on new calibration points, thereby achieving more stable and accurate convergence.

[0029] The current analysis model is obtained by fitting each calibration point in the current curve fitting dataset. The current curve fitting dataset is set as a weighted set of (OCV, SOC, weight) triplet points, and a cubic B-spline is used for fitting. U nivariateSpline (OCV, SOC, k=3, s=ad_smooth, w=weight) Where k=3 represents a cubic B-spline, ensuring the continuity of the second derivative; s is the adaptive smoothing factor; and w is the weight vector of each calibration point.

[0030] Advantages of B-splines over piecewise polynomials: Cubic B-splines can achieve the fitting ability of higher-order polynomials, are naturally piecewise without needing to explicitly specify the piecewise points, do not exhibit Runge's phenomenon, and have high fitting accuracy and good numerical stability.

[0031] This embodiment does not require prior knowledge of the precise SOC-OCV curve of the entire cluster, nor does it require disassembly for offline calibration or complete offline charge and discharge. It is a self-supervised method that can estimate the actual capacity of each individual cell online using actual operating data of the energy storage system. It can adapt to battery aging and individual cell differences, and provides a reliable basis for capacity calculation through accurate SOC calibration, thereby improving the capacity estimation accuracy and individual cell consistency diagnostic capability of the energy storage battery system.

[0032] In some embodiments, S110: Before acquiring the actual operating data of the energy storage battery system at the current moment, the method further includes: S210: Obtain the initial operating data of the energy storage battery system at the initial moment, where the initial moment is the first moment when the generative analysis model is started; S220: Align the initial running data with a reference, and obtain the initial calibration point by combining it with a preset prior curve; S230: Fit the initial calibration points to obtain the initial analysis model.

[0033] S220: Align the initial running data with a reference, and obtain the initial calibration point by combining it with a preset prior curve, including: S221: Perform reference alignment on the initial operating data to determine the battery information at the discharge cutoff time and the charging cutoff time, wherein the battery information includes open circuit voltage information and charge information; S222: Obtain prior data points by integrating the initial running data and the preset prior curve; S223: Determine the initial calibration point based on the battery information at the discharge cutoff time and the charging cutoff time, and the prior data points.

[0034] Specifically, the current analysis model at each current moment is obtained by iterating from the previous analysis model. It should be noted that the previous analysis model is the model that was most recently iterated relative to the current analysis model. The actual time interval between the two is uncertain and depends on the iteration conditions and iteration cycle of the model.

[0035] The initial time is the first moment when the analysis model is generated, that is, when the analysis model is generated at the initial time, there is no previous analysis model of the initial analysis model.

[0036] Acquire initial operating data of the energy storage battery system at the initial moment, perform benchmark alignment on the initial operating data, and determine the battery information at the discharge cutoff moment and the charging cutoff moment. The battery information includes open circuit voltage information and charge information.

[0037] The specific methods for datum alignment include: Discharge cutoff time (after the entire cluster has finished discharging and is left to stand): Find the cell with the lowest voltage among all cells, and calibrate its SOC to the preset discharge cutoff reference value SOCdis_ref (range 0~0.15), as the absolute calibration point, with a weight of 1.0; calculate the initial SOC of the remaining cells according to the preset slope: SOC = SOCdis_ref + (OCV - OCV_min)×slope_low, where slope_low = 1 / DVDSOC_LOW is the reciprocal of the preset SOC-OCV slope in the later stage of discharge, obtained from the preset SOC-OCV curve. OCV is the voltage at the discharge cutoff time of the corresponding cell, and OCV_min is the cell with the lowest voltage at the discharge cutoff time among all cells.

[0038] Charging cutoff time: The cell with the highest voltage among all cells has its calibrated SOC set as the preset charging cutoff reference value SOCch_ref (range 0.90~1.0), which serves as the absolute calibration point with a weight of 1.0; the initial SOC of the remaining cells is calculated according to a preset slope. SOC = SOCch_ref - (OCV_max - OCV) × slope_high, where slope_high = 1 / DVDSOC_HIGH, which is the inverse of the preset SOC-OCV slope in the later stages of charging. OCV is the voltage at the charging cutoff point of the corresponding cell, and OCV_max is the cell with the highest voltage at the charging cutoff point among all cells.

[0039] Furthermore, prior data points are obtained based on the initial running data and the integration of a preset prior curve. A set of (SOC, OCV) prior data points is obtained by integrating the preset prior curve dSOC / dV data, and a prior point weight w_prior (range 0.05~0.5) is assigned. This prior point is added to the input dataset of the current curve fitting dataset as weak prior information to assist in platform area fitting. Forced charge / discharge cutoff boundary: If OCV≤OCV_min, then SOC = SOCdis_ref is forced; if OCV≥OCV_max, then SOC = SOCch_ref is forced. The prior point weight w_prior is used to configure the confidence level of the SOC-OCV calibration points obtained from the prior curve integration in B-spline fitting, with a value range of 0.05~0.5.

[0040] Prior point weight configuration strategy: In the first round of initial calibration point collection, the SOC-OCV points obtained by prior curve integration are assigned a weight w_prior (default value 0.1). The discharge cutoff point and the charge cutoff point are used as actual measurement calibration points and are assigned a full weight of 1.0. When the prior quality is good, w_prior can be increased to 0.3; when the prior quality is poor, w_prior can be decreased to 0.05. The weight of prior points is lower than that of measured points to ensure that the algorithm prioritizes the measured data, while using prior information to provide platform area estimation support.

[0041] Furthermore, for the lithium iron phosphate battery system, the prior curve was obtained by normalizing the historical dSOC / dV data after Savitzky-Golay smoothing; for the ternary lithium battery system, the same method was used but the smoothing parameters were adjusted.

[0042] An initial analysis model is obtained by fitting battery information and prior data points at the discharge and charge cutoff times. Specifically, the initial calibration points include the prior curve integration points (weight w_prior=0.1), the discharge cutoff point (weight 1.0), and the charge cutoff point (weight 1.0). A cubic B-spline (k=3) is used for fitting, with an adaptive smoothing factor s=len(x)*var(y)*0.01. The SOC is directly fitted as a function of OCV: SOC = f(OCV). The next iteration is based on the initial analysis model to obtain the next analysis model, and the iteration after that is based on the next analysis model, and so on.

[0043] This embodiment introduces a weighted B-spline fitting mechanism. By configuring prior point weights and old point weight decay factors, it achieves differentiated processing of new and old calibration points. This allows the algorithm to fully utilize historical data during iteration while prioritizing the newly generated calibration points, thereby significantly improving the SOC-OCV function fitting accuracy and convergence speed, and providing an accurate SOC basis for subsequent capacity estimation of each individual cell.

[0044] In some embodiments, step S221: performing reference alignment on the initial operating data to determine battery information at the discharge cutoff time and the charging cutoff time includes: S310: Determine the voltage data, charge data, and total cluster current data of each cell based on the initial operating data; S320: Identify the charge / discharge cycle based on the total cluster current data; S330: Identify at least two complete charge-discharge cycles and determine the open-circuit voltage of each cell at the charge-discharge cutoff time based on the voltage data of each cell; S340: Determine the calibration reference period from the at least two complete charge-discharge cycles based on the open-circuit voltage of each cell at the charge-discharge cutoff time; S350: Align the voltage data of each cell, the charge data, and the charge / discharge cycle with a reference to determine the battery information at the discharge cutoff time and the charging cutoff time.

[0045] Specifically, based on the initial operating data, the voltage data, charge data, and total cluster current data of each cell are determined, and the charge / discharge cycle is identified based on the total cluster current data. At least two complete charge / discharge cycles are identified for dual-cycle cross-calibration. Dual-cycle cross-calibration specifically includes: comparing the cutoff voltages of at least two discharge cycles, selecting the cycle with the lower discharge cutoff voltage (i.e., more complete discharge) as the SOC-OCV calibration reference cycle; using the cell calibration point SOC=SOCdis_ref with the lowest discharge cutoff voltage and the cell calibration point SOC=SOCch_ref with the highest charge cutoff voltage of this calibration cycle, along with the initial SOC of the remaining cells calculated according to a preset slope, and combining the data points obtained by integrating the prior SOC-OCV curve, a cubic B-spline UnivariateSpline (k=3, s=adaptive) is used for fitting. The input is a weighted set of (OCV, SOC, weight) triplet points, and the output is the SOC=f(OCV) mapping function.

[0046] In some embodiments, step S150: if the current analysis model meets the preset convergence condition, then analyze the battery capacity of the energy storage battery system based on the current analysis model, including: S410: If the current analysis model meets the preset convergence conditions, then the cumulative charge of the discharge process and the charge change at the start and end of the discharge are determined based on the current analysis model. S420: Determine the battery capacity based on the cumulative charge and the charge change.

[0047] Specifically, the preset convergence condition is that the average capacity change of all cells is less than the preset convergence threshold epsilon, where epsilon ranges from 0.0001 to 0.005. In the first iteration, the lowest voltage cell is forced to have SOC = SOCdis_ref, and the highest voltage cell is forced to have SOC = SOCch_ref as absolute calibration points. In subsequent iterations, SOC is only clamped to the range [0, 1] to prevent out-of-bounds errors, and boundary conditions are no longer enforced. The current analysis model can be a new iteration where the previous analysis model did not meet the preset convergence condition, or an iteration update where the previous analysis model met the preset convergence condition and the iteration cycle has ended.

[0048] If the current analysis model meets the preset convergence conditions, the trapezoidal integral is used to calculate the accumulated charge during the discharge process, and the change in SOC is calculated based on the SOC at the beginning and end of the discharge. Battery capacity = accumulated charge / change in SOC.

[0049] This embodiment utilizes the actual charging and discharging data of the energy storage system, performs self-supervised calibration based on the prior knowledge of cell voltage consistency, and obtains accurate SOC-OCV curves and cell capacities through multiple rounds of iterative optimization. It does not require disassembly for offline calibration and is adaptive to different battery systems and aging levels.

[0050] This application provides a method for analyzing the capacity of an energy storage battery in some embodiments. Taking a certain energy storage system with 400 lithium iron phosphate cells connected in series as an example, the specific implementation steps are as follows: Step S100: Data Acquisition and Preprocessing Acquire actual operating data of the energy storage battery system, identify at least two complete charge-discharge cycles, and extract the open-circuit voltage (OCV) of each cell at the charge-discharge cutoff moment.

[0051] Input data: voltage data of each cell (sampling frequency 1 minute / time), total current data of the cluster (sampling frequency 1 second / time). - Charge / discharge cycle identification: Identify charge / discharge events based on changes in current direction, filter short cycles, and select the two longest charge / discharge cycles; -OCV extraction: After the charging and discharging is cut off, the average value of the last 5 voltage points within the last 30 minutes of rest is taken as the OCV.

[0052] In this embodiment, the current identification threshold Ith is set to 5A, the minimum cycle duration Tmin is 5 minutes, the OCV time window is 30 minutes, and the average of 5 points is taken.

[0053] Step S200: Baseline Alignment and Initial SOC Calculation Dual-cycle cross-calibration: Compare the cutoff voltage of at least two discharge cycles, select the cycle with more complete discharge as the SOC-OCV calibration reference cycle, and use the calibration point SOC of this calibration cycle and the initial SOC calculated from the remaining cells.

[0054] The specific strategy in this embodiment is as follows: Discharge cutoff time (after the entire cluster discharges and is left to stand): Find the cell with the lowest voltage among all cells and calibrate its SOC = SOCdis_ref (a preset discharge cutoff reference value, ranging from 0 to 0.15, and 0.02 in this embodiment), with a weight of 1.0; calculate the initial SOC of the remaining cells according to the preset slope = SOCdis_ref + (OCV - OCV_min) × slope_low, with a weight of 1.0.

[0055] Charging cutoff time (after the entire cluster is charged and left to stand): Find the cell with the highest voltage among all cells and calibrate its SOC = SOCch_ref (a preset charging cutoff reference value, ranging from 0.90 to 1.0, and 0.97 in this embodiment), with a weight of 1.0; calculate the initial SOC of the remaining cells according to the preset slope = SOCch_ref-(OCV_max - OCV)×slope_high, with a weight of 1.0.

[0056] In addition, the (SOC, OCV) data points obtained by integrating the prior dSOC / dV curve are added to the B-spline fitting input dataset with a prior point weight w_prior=0.1 as weak prior information to improve the fitting accuracy of the plateau region.

[0057] Step S300: Weighted B-spline fitting of SOC (OCV) Using the calibration point SOC of the calibration cycle and the initial SOC calculated from the remaining cells, combined with the (SOC, OCV) point obtained by integrating the prior dSOC / dV curve (with prior point weight w_prior, ranging from 0.05 to 0.5), the SOC(OCV) function model is fitted using a weighted B-spline fitting method.

[0058] - Collect initial calibration points: prior curve integration point (weight w_prior=0.1), discharge cutoff point (weight 1.0), and charging cutoff point (weight 1.0). - A cubic B-spline (k=3) was used for fitting, with an adaptive smoothing factor s=len(x)*var(y)*0.01. A schematic diagram of the weighted B-spline fitting is shown below. Figure 2 As shown; - Directly fit SOC as a function of OCV: SOC = f(OCV).

[0059] Step S400: Multi-round iterative optimization (including old point weight decay) First round: Use the initial calibration points obtained by benchmark alignment and the weak prior data points obtained by prior curve integration to perform B-spline fitting to obtain the initial SOC (OCV) model; Subsequent rounds: (1) Recalculate the SOC of all charge and discharge cutoff points using the current B-spline model, clamping it to the range of [0, 1]; (2) The new calibration points are added to the dataset with a full weight of 1.0; (3) The historical calibration point weights are multiplied by the decay factor alpha=0.99 for exponential decay; (4) Update the dataset and re-fit the weighted B-spline to obtain a new model; (5) Calculate the average capacity change; (6) If the average capacity change is less than epsilon (0.001 in this example), convergence stops; otherwise, iteration continues.

[0060] In this embodiment, the old point weight decay factor alpha = 0.99, and convergence is generally achieved in 5 to 8 rounds.

[0061] Step S500: Capacity Calculation - Use trapezoidal integral to calculate the cumulative charge during the discharge process; - Calculate the change in SOC based on the SOC at the start and end of the discharge; - Capacity = Accumulated charge / Change in SOC; - Statistically analyze the capacity distribution of all battery cells and calculate statistical parameters.

[0062] The present invention also provides an energy storage battery capacity analysis device for implementing the above method, comprising: Data acquisition module: used to collect voltage data of each cell and total current data of the energy storage battery system, identify the charge and discharge cycle, and extract the open circuit voltage (OCV) of each cell at the charge and discharge cutoff time; The benchmark alignment module is used to perform benchmark SOC calibration on cells with the lowest discharge cutoff voltage and the highest charge cutoff voltage based on voltage consistency priors. The initial SOC of the remaining cells is calculated according to a preset slope. The integral points of the prior dSOC / dV curve are added as weighted data points to the subsequent B-spline fitting dataset. Weighted B-spline fitting module: Used to fit the SOC-OCV calibration points using the weighted B-spline fitting method, configure the prior point weight w_prior and the old point weight decay factor alpha, and directly obtain the SOC(OCV) function model; Iterative optimization module: used to continuously update the SOC-OCV dataset through multiple rounds of iteration. New calibration points are added with full weight, and historical calibration points are exponentially decayed according to the decay factor. The refit optimization continues until convergence. Capacity Calculation Module: Used to calculate the SOC change of each cell based on the converged SOC model, calculate the actual capacity of each cell by combining the integrated charge, and output the statistical results.

[0063] Figure 3 This invention illustrates a schematic diagram of a battery capacity analysis device 900 provided in some embodiments of the present application. The device includes: Data acquisition module 910 is used to acquire the actual operating data of the energy storage battery system at the current moment; The data analysis module 920 is used to obtain the current calibration point based on the actual operating data and the previous analysis model. The current calibration point is a SOC-OCV data point. The previous analysis model includes the previous curve fitting dataset. The curve fitting dataset includes several historical calibration points and the historical weights of each historical calibration point. The model analysis module 930 is used to add the current calibration point and its corresponding calibration point weights to the previous curve fitting dataset, and update the historical weights of each historical calibration point in the previous curve fitting dataset according to a preset weight decay factor to obtain the current curve fitting dataset; and to fit each calibration point in the current curve fitting dataset to obtain the current analysis model, wherein the previous analysis model is obtained by fitting each calibration point in the previous curve fitting dataset. The capacity analysis module 940 is used to analyze the battery capacity of the energy storage battery system based on the current analysis model if the current analysis model meets the preset convergence conditions.

[0064] In some embodiments, the model analysis module 930 is further configured to acquire initial operating data of the energy storage battery system at an initial moment, wherein the initial moment is the first moment when the generation of the analysis model is started; perform benchmark alignment on the initial operating data and combine it with a preset prior curve to obtain an initial calibration point; and fit the initial calibration point to obtain an initial analysis model.

[0065] In some embodiments, the model analysis module 930 is further configured to perform benchmark alignment on the initial operating data, determine battery information at the discharge cutoff time and the charging cutoff time, the battery information including open circuit voltage information and charge information; obtain prior data points by integrating the initial operating data and a preset prior curve; and determine the initial calibration point based on the battery information at the discharge cutoff time and the charging cutoff time and the prior data points.

[0066] In some embodiments, the model analysis module 930 is further configured to determine the voltage data, charge data, and total cluster current data of each cell based on the initial operating data; identify the charge / discharge cycle based on the total cluster current data; perform reference alignment on the voltage data of each cell, the charge data, and the charge / discharge cycle to determine the battery information at the discharge cutoff time and the charging cutoff time.

[0067] In some embodiments, the model analysis module 930 is further configured to identify at least two complete charge-discharge cycles, determine the open-circuit voltage of each cell at the charge-discharge cutoff time based on the cell voltage data, determine a calibration reference period from the at least two complete charge-discharge cycles based on the open-circuit voltage of each cell at the charge-discharge cutoff time, and perform reference alignment on the cell voltage data, the charge data, and the charge-discharge cycle within the calibration reference period to determine the battery information at the discharge cutoff time and the charge cutoff time.

[0068] In some embodiments, the preset convergence condition is set to the average capacity change being less than a preset threshold.

[0069] In some embodiments, the capacity analysis module 940 is further configured to determine the cumulative charge during the discharge process and the charge change at the start and end of the discharge based on the current analysis model if the current analysis model meets the preset convergence conditions; and to determine the battery capacity based on the cumulative charge and the charge change.

[0070] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0071] Accordingly, embodiments of this application also provide an electronic device, which includes one or more processors; a memory; and one or more application programs, wherein the one or more application programs are stored in the memory and configured to be executed by the processors to perform the steps of the above-described energy storage battery capacity analysis method.

[0072] In some embodiments of this application, a computer-readable storage medium is provided, which stores a computer program that is loaded by a processor to cause the processor to execute the steps of the above-described energy storage battery capacity analysis method.

[0073] This application does not limit the type of processor. A processor is a circuit with signal processing capabilities. For example, a processor includes circuits with instruction read and execute capabilities, such as a CPU, microprocessor, graphics processing unit (GPU), or digital signal processor (DSP). Alternatively, a processor can be implemented through the logical relationships of hardware circuits, where these relationships are fixed or reconfigurable. For instance, a processor may include hardware circuits implemented using an ASIC or PLD, such as a field-programmable gate array (FPGA). In reconfigurable hardware circuits, the process of the processor loading a configuration document and configuring the hardware circuits can be understood as the processor loading instructions to implement the functions of some or all of the aforementioned units. Furthermore, a processor may include hardware circuits designed for artificial intelligence, which can be understood as an ASIC, such as one or more of a neural network processing unit (NPU), tensor processing unit (TPU), or deep learning processing unit (DPU).

[0074] As can be seen, each unit in the above energy storage battery capacity analysis device can be one or more processors (or processing circuits) configured to implement the above method, such as: CPU, GPU, NPU, TPU, DPU, microprocessor, DSP, ASIC, FPGA, or a combination of at least two of these processor forms.

[0075] Furthermore, the units in the above-mentioned energy storage battery capacity analysis device can be integrated in whole or in part, or they can be implemented independently. In one implementation, these units are integrated together as a system-on-a-chip (SOC). The SOC may include at least one processor for implementing any of the above-mentioned energy storage battery capacity analysis methods or for implementing the functions of each unit in the above-mentioned chip testing device.

[0076] This application does not limit the type of memory. For example, the memory may include read-only memory (ROM) or random access memory (RAM). It may include non-volatile or volatile memory. It may include on-chip or off-chip storage resources. ROM includes, for example, mask ROM, programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash ROM; RAM includes, for example, static random access memory (SRAM) or dynamic random access memory (DRAM). On-chip storage resources refer to the storage circuitry integrated within the chip. This circuitry is integrated with other circuitry (such as processing circuitry) to provide storage space for the chip. Examples include on-chip static random access memory (SRAM) resources or on-chip configuration register space resources. Off-chip storage resources refer to independent storage circuitry or chips, such as off-chip double data rate synchronous dynamic RAM (DDRSDRAM, or DDR for short).

[0077] In this disclosure, unless otherwise expressly specified and limited, ordinal numbers, such as "first," "second," etc., are used only to distinguish and describe related objects, and should not be construed as indicating or implying the relative importance or order between related objects. Furthermore, ordinal numbers do not represent the quantity of related objects.

[0078] "Multiple" includes two or more, and other classifiers are similar.

[0079] The terms "or" and "and / or" in this disclosure are used to describe relationships between related objects, indicating a non-exclusive inclusion. For example, "A and / or B" and "A or B" can both include: "A alone," "B alone," or "A and B," where "A" and "B" can include a single object or multiple objects. Similarly, "A, B and / or C," "A, B or C," and "A, B and C" can both include: "A alone," "B alone," "C alone," "A and B," "A and C," "B and C," or "A, B and C," where "A," "B," and "C" can include a single object or multiple objects. Additionally, the " / " in this disclosure is used to indicate an "or" relationship between related objects. The meanings of "at least one of A or B" and "one or more of A and B" in this disclosure are the same as the meaning of "A or B" above. The meanings of "one or more of A, B, and C" and "at least one of A, B, or C" are the same as the meaning of "A, B, or C" above. The meaning of "one or more of A, B, and C" is the same as the meaning of "A, B, or C" above.

[0080] In the above embodiments, the descriptions of each embodiment have their own emphasis. Parts not described in detail or in a particular embodiment can be referred to in the relevant descriptions of other embodiments. Furthermore, the above embodiments can be freely combined as needed.

Claims

1. A method for analyzing the capacity of an energy storage battery, characterized in that, The method includes: Obtain the current operating data of the energy storage battery system; The current calibration point is obtained based on the actual operating data and the previous analysis model. The current calibration point is a SOC-OCV data point. The previous analysis model includes the previous curve fitting dataset. The curve fitting dataset includes several historical calibration points and the historical weights of each historical calibration point. The current calibration point and its corresponding calibration point weights are added to the previous curve fitting dataset, and the historical weights of each historical calibration point in the previous curve fitting dataset are updated according to the preset weight decay factor to obtain the current curve fitting dataset. The current analysis model is obtained by fitting each calibration point in the current curve fitting dataset, and the previous analysis model is obtained by fitting each calibration point in the previous curve fitting dataset. If the current analysis model meets the preset convergence conditions, then the battery capacity of the energy storage battery system is analyzed based on the current analysis model.

2. The energy storage battery capacity analysis method according to claim 1, characterized in that, Before obtaining the actual operating data of the energy storage battery system at the current moment, the following steps are also included: Acquire the initial operating data of the energy storage battery system at the initial moment, where the initial moment is the first moment when the generative analysis model is started; The initial running data is aligned to a reference, and the initial calibration point is obtained by combining it with a preset prior curve. The initial analysis model is obtained by fitting the initial calibration points.

3. The energy storage battery capacity analysis method according to claim 2, characterized in that, The step involves aligning the initial running data to a reference point and, in conjunction with a preset prior curve, obtaining initial calibration points, including: The initial operating data is aligned to a reference to determine the battery information at the discharge cutoff time and the charging cutoff time. The battery information includes open-circuit voltage information and charge information. Prior data points are obtained by integrating the initial running data and the preset prior curve; The initial calibration point is determined based on the battery information at the discharge cutoff time and the charging cutoff time, and the prior data points.

4. The energy storage battery capacity analysis method according to claim 3, characterized in that, The step involves aligning the initial operating data to a reference, wherein the initial operating data determines battery information at the discharge cutoff time and the charge cutoff time, including: Based on the initial operating data, determine the voltage data, charge data, and total cluster current data of each cell; The charge / discharge cycle is identified based on the total cluster current data; The voltage data of each cell, the charge data, and the charge / discharge cycle are aligned to a reference to determine the battery information at the discharge cutoff time and the charging cutoff time.

5. The energy storage battery capacity analysis method according to claim 4, characterized in that, Before the step of aligning the voltage data of each cell, the charge data, and the charge / discharge cycle with a reference to determine the battery information at the discharge cutoff time and the charge cutoff time, it also includes: Identify at least two complete charge-discharge cycles, and determine the open-circuit voltage of each cell at the charge-discharge cutoff time based on the voltage data of each cell; The calibration reference period is determined from the at least two complete charge-discharge cycles based on the open-circuit voltage of each cell at the charge-discharge cutoff time. The step involves aligning the voltage data of each cell, the charge data, and the charge / discharge cycle to a reference, and determining the battery information at the discharge cutoff time and the charge cutoff time, including: The voltage data of each cell, the charge data, and the charge / discharge cycle within the calibration reference period are aligned to determine the battery information at the discharge cutoff time and the charge cutoff time.

6. The energy storage battery capacity analysis method according to claim 1, characterized in that, The preset convergence condition is set to the average capacity change being less than a preset threshold.

7. The energy storage battery capacity analysis method according to any one of claims 1-6, characterized in that, If the current analysis model meets the preset convergence conditions, then the battery capacity of the energy storage battery system is analyzed based on the current analysis model, including: If the current analysis model meets the preset convergence conditions, then the cumulative charge during the discharge process, as well as the charge change at the start and end of the discharge, are determined based on the current analysis model. The battery capacity is determined based on the cumulative charge and the change in charge.

8. A battery capacity analysis device, characterized in that, The device includes: The data acquisition module is used to acquire the actual operating data of the energy storage battery system at the current moment; The data analysis module is used to obtain the current calibration point based on the actual operating data and the previous analysis model. The current calibration point is a SOC-OCV data point. The previous analysis model includes the previous curve fitting dataset. The curve fitting dataset includes several historical calibration points and the historical weights of each historical calibration point. The model analysis module is used to add the current calibration point and its corresponding calibration point weights to the previous curve fitting dataset, and update the historical weights of each historical calibration point in the previous curve fitting dataset according to a preset weight decay factor to obtain the current curve fitting dataset; and to fit each calibration point in the current curve fitting dataset to obtain the current analysis model, wherein the previous analysis model is obtained by fitting each calibration point in the previous curve fitting dataset. The capacity analysis module is used to analyze the battery capacity of the energy storage battery system based on the current analysis model if the current analysis model meets the preset convergence conditions.

9. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the energy storage battery capacity analysis method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a computer program, which, when executed, controls the electronic device containing the computer-readable storage medium to perform the energy storage battery capacity analysis method according to any one of claims 1 to 7.

Images