Methods, devices, storage media, and processors for determining battery capacity

Abstract

This invention discloses a method, apparatus, storage medium, and processor for determining battery capacity. The method includes: acquiring current and voltage data of the battery; inputting the current and voltage data into a capacity estimation model for processing to obtain health indicators of the battery under different aging states, wherein the capacity estimation model is used to determine the performance data of different batteries under different aging states; determining the estimated capacity of the battery based on the health indicators of the battery under different aging states and the correlation function between the health indicators and the actual capacity of the battery, wherein the correlation function is used to characterize the linear relationship between the health indicators and the actual capacity; and performing error analysis on the estimated capacity of the battery based on the actual capacity to determine the estimated battery capacity result. This invention solves the technical problem of low accuracy in battery capacity prediction.

Description

Technical Field

[0001] This invention relates to the field of information processing technology, and more specifically, to a method, apparatus, storage medium, and processor for determining battery capacity. Background Technology

[0002] To address the increasingly prominent contradiction between fuel supply and demand and environmental pollution, electric vehicles (EVs) have received widespread attention both domestically and internationally due to their non-reliance on fossil fuels and zero emissions during operation. As the power source of EVs, the power battery plays a crucial role in their performance. Power batteries can be categorized based on electrode materials into lead-acid batteries, nickel-cadmium batteries, nickel-metal hydride batteries, and lithium-ion batteries. Among these, lithium-ion batteries dominate the EV industry due to their high energy density, long cycle life, wide temperature adaptability, and low self-discharge rate. To extend battery life and ensure the reliability of EVs, accurately estimating the battery's state-of-health (SOH) is essential. Currently, battery SOH is primarily defined as the ratio of actual usable capacity to initial usable capacity. Generally, the initial capacity is a known fixed value; therefore, estimating the battery SOH is equivalent to estimating the current actual usable capacity of the power battery. Because power batteries gradually age during use, some active lithium ions are lost due to irreversible chemical reactions; therefore, the battery SOH generally decreases with battery use. Because battery SoH (Power Output) cannot be directly measured by sensors such as voltage and current, and because the usage scenarios of electric vehicles are highly dynamic and complex, accurately estimating battery SoH becomes extremely difficult. Inaccurate SoH estimation not only compromises the safety and reliability of the power battery, but also, due to the coupling relationship between SoH and the remaining battery capacity (State of Charge, or SOC), severely impacts the accuracy of SOC estimation.

[0003] Currently, battery SoH estimation methods can be divided into three categories: (1) electrochemical methods; (2) equivalent circuit methods; and (3) data-driven methods. Among them, the electrochemical method is mainly based on the aging mechanism of the battery itself and has the best estimation accuracy among the three methods. However, the model involves too many parameters, and the calibration and updating of it are cumbersome, making it difficult to apply in battery management systems (BMS) with limited computing resources. To this end, the equivalent circuit method with fewer model parameters has been proposed. This method is popular because of its good balance between computational burden and model accuracy. However, it relies on online parameter identification algorithms such as recursive least squares algorithm to update model parameters, and the robustness of the model still faces great challenges. In addition, this method is usually combined with filtering algorithms such as particle filtering or Kalman filtering. If the prior probability distribution or noise variance is not properly predefined in these filtering algorithms, the divergence of the method cannot be guaranteed.

[0004] To overcome the shortcomings of electrochemical and equivalent circuit methods, a data-driven approach is proposed. This method, by mining large amounts of data, can directly analyze the intrinsic relationship between battery SoH and its related influencing factors (commonly referred to as health indicators). Its key advantage is that it does not require prior in-depth understanding of complex battery aging mechanisms. Furthermore, since the model is calibrated during training, its computational burden during application is relatively light. The selection of health indicators has a significant impact on model estimation performance. To date, many health indicators have been extracted from battery current, voltage, etc. Although these health indicators have a certain correlation with battery degradation, their relationship is non-linear. Non-linear models such as Support Vector Machines (SVM) or Grey Relational Analysis (GRA) are used to map health indicators to battery capacity. However, due to the non-linear nature, which is to some extent uninterpretable and unstable, the robustness of the estimated battery SoH may decrease when the non-linear model is incorrectly calibrated or features are incorrectly extracted or calculated, leading to lower accuracy in battery capacity estimation.

[0005] There is currently no effective solution to the technical problem of low accuracy in battery capacity prediction. Summary of the Invention

[0006] This invention provides a method, apparatus, storage medium, and processor for determining battery capacity, to at least address the technical problem of low accuracy in battery capacity estimation.

[0007] According to one aspect of the present invention, a method for determining battery capacity is provided. The method may include: acquiring current data and voltage data of the battery; inputting the current data and voltage data into a capacity estimation model for processing to obtain health indicators of the battery under different aging states, wherein the capacity estimation model is used to determine the performance data of different batteries under different aging states; determining the estimated capacity of the battery based on the health indicators of the battery under different aging states and a correlation function between the health indicators and the actual capacity of the battery, wherein the correlation function is used to characterize the linear relationship between the health indicators and the actual capacity; and performing error analysis on the estimated capacity of the battery based on the actual capacity to determine the estimated battery capacity result.

[0008] Optionally, the current and voltage data are input into the capacity estimation model for processing to obtain the battery's health indicators under different aging states, including: performing spectral entropy analysis on the current and voltage data to obtain frequency domain features; performing information energy analysis on the frequency domain features to obtain an energy information matrix; and performing singular value decomposition on the energy information matrix to obtain the battery's health indicators under different aging states.

[0009] Optionally, singular value decomposition is performed on the energy information matrix to obtain the battery health indicators under different aging states, including: performing singular value decomposition on the energy information matrix to obtain at least one sub-matrix and the singular values ​​corresponding to each sub-matrix; using the singular values ​​corresponding to each sub-matrix to extract signal features in the energy information matrix; and determining the battery health indicators based on the signal features in the energy information matrix.

[0010] Optionally, before inputting the battery's current and voltage data into the capacity estimation model for processing to obtain the battery's health indicators under different aging states, the method further includes: extracting features from the current and voltage data samples of the battery samples; determining the health indicators of the battery samples based on the extracted features; performing linear regression between the health indicators of the battery samples and the calibrated capacity to obtain the regression results; and establishing a capacity estimation model based on the current data samples, voltage data samples, and regression results.

[0011] Optionally, error analysis is performed on the estimated capacity of the battery based on the actual capacity to obtain the estimated battery capacity result, including: obtaining the relative error and average error between the actual capacity and the estimated capacity; in response to the relative error being less than or equal to a relative error threshold and the average error being less than or equal to an average error threshold, the estimation result is determined, wherein the relative error threshold is a preset condition for evaluating whether the relative error is reasonable and the average error threshold is a preset condition for evaluating whether the average error is reasonable.

[0012] Optionally, the method further includes: in response to the relative error being greater than a relative error threshold, and / or the average error being greater than an average error threshold, performing multiple linear regressions on the initial capacity estimation model using the relative error and the average error until the initial capacity estimation model converges; and determining the converged initial capacity estimation model as the capacity estimation model.

[0013] According to another aspect of the present invention, a battery capacity determination apparatus is also provided, comprising: an acquisition unit for acquiring current data and voltage data in a battery; a processing unit for inputting the current data and voltage data into a capacity estimation model for processing to obtain health indicators of the battery under different aging states, wherein the capacity estimation model is used to determine the characteristics of the battery under different aging states; a determination unit for determining the estimated capacity of the battery based on the health indicators of the battery under different aging states and a correlation function between the health indicators and the actual capacity of the battery, wherein the correlation function is used to characterize the linear relationship between the health indicators and the actual capacity; and an analysis unit for performing error analysis on the estimated capacity of the battery based on the actual capacity to determine the estimated result of the battery capacity.

[0014] According to another aspect of the present invention, a computer-readable storage medium is also provided. The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to perform the battery capacity determination method of the present invention.

[0015] According to another aspect of the present invention, a processor is also provided. The processor is used to run a program, wherein the program executes the battery capacity determination method of the present invention during runtime.

[0016] According to another aspect of the present invention, a vehicle is also provided for performing the battery capacity determination method of the present invention.

[0017] In this embodiment of the invention, current and voltage data of the battery are acquired; the current and voltage data are input into a capacity estimation model for processing to obtain the battery's health indicators under different aging states. The capacity estimation model is used to determine the performance data of different batteries under different aging states. Based on the battery's health indicators under different aging states and the correlation function between the health indicators and the battery's actual capacity, the estimated capacity of the battery is determined. The correlation function characterizes the linear relationship between the health indicators and the actual capacity. Error analysis is performed on the estimated capacity of the battery based on the actual capacity to determine the estimated battery capacity result. In other words, in this embodiment of the invention, the battery's health indicators under different aging degrees can be obtained first based on the capacity estimation model. Then, based on the correlation function between the health indicators and the battery's actual capacity, the estimated capacity of the battery can be determined. Since the correlation function is the correlation function between the battery's health indicator parameters and the actual capacity, the battery capacity determined based on this correlation function is relatively accurate, achieving the goal of improving the robustness of battery capacity estimation. Finally, error analysis can be performed on the estimated capacity of the battery to further improve the accuracy of battery capacity prediction, achieving the technical effect of improving the accuracy of battery capacity estimation and solving the technical problem of low accuracy in battery capacity prediction. Attached Figure Description

[0018] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0019] Figure 1 This is a flowchart of a method for determining battery capacity according to an embodiment of the present invention;

[0020] Figure 2 This is a schematic diagram of a battery capacity estimation method according to an embodiment of the present invention;

[0021] Figure 3 This is a flowchart of another method for estimating battery capacity according to an embodiment of the present invention;

[0022] Figure 4 This is a schematic diagram illustrating how a first health indicator changes with battery capacity according to an embodiment of the present invention;

[0023] Figure 5 This is a schematic diagram illustrating the change in battery capacity according to a second health indicator based on an embodiment of the present invention.

[0024] Figure 6 This is a schematic diagram of the estimated capacity and the actual capacity according to an embodiment of the capacity estimation method of the present invention;

[0025] Figure 7 This is a schematic diagram illustrating the relative error and average error between the estimated capacity and the actual capacity of a capacity estimation method according to an embodiment of the present invention.

[0026] Figure 8 This is a schematic diagram of a battery capacity determination device according to an embodiment of the present invention. Detailed Implementation

[0027] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0028] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0029] Example 1

[0030] According to an embodiment of the present invention, an embodiment of a method for determining battery capacity is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0031] Figure 1 This is a flowchart of a method for determining battery capacity according to an embodiment of the present invention, such as... Figure 1 As shown, the method may include the following steps:

[0032] Step S101: Obtain the battery's current and voltage data.

[0033] In the technical solution provided by step S101 of the present invention, the current data and voltage data of batteries under different aging states can be obtained by experimentally measuring them. The experimental method is not specifically limited here.

[0034] Step S102: Input the current data and voltage data into the capacity estimation model for processing to obtain the health indicators of the battery under different aging conditions. The capacity estimation model is used to determine the performance data of different batteries under different aging conditions.

[0035] In the technical solution provided in step S102 of the present invention, the capacity estimation model is used to determine the performance data of different batteries under different aging states. Based on this, after obtaining the current data and voltage data under different aging states, the obtained current data and voltage data can be input into the capacity estimation model for processing to obtain the health indicators of the battery under different aging states.

[0036] Optionally, after receiving battery current and voltage data under different aging conditions, the capacity estimation model can first perform spectral entropy analysis on the current and voltage data to extract the frequency domain features of the current and voltage signals. Then, it can perform information energy analysis on the frequency domain features to obtain an energy information matrix, and perform singular value decomposition on the energy information matrix to obtain the battery health indicators under different aging conditions.

[0037] Step S103: Based on the health indicators of the battery under different aging states and the correlation function between the health indicators and the actual capacity of the battery, determine the estimated capacity of the battery, wherein the correlation function is used to characterize the linear relationship between the health indicators and the actual capacity.

[0038] In the technical solution provided by step S103 of the present invention, after determining the health indicators of the battery under different aging states, the battery capacity can be estimated based on the correlation function between the battery health indicators and the battery's actual capacity. The correlation function is used to characterize the linear relationship between the health indicators and the actual capacity.

[0039] Optionally, after determining the health indicators of the battery under different aging conditions, the health indicators of the battery under different aging conditions can be input into the correlation function for calculation to obtain the battery capacity under the corresponding aging conditions.

[0040] Step S104: Perform error analysis on the estimated capacity of the battery based on the actual capacity to determine the estimated battery capacity.

[0041] In the technical solution provided by step S104 of the present invention, after determining the estimated capacity of the battery under different aging states based on the correlation function, the estimated battery capacity under different aging states can be analyzed for error based on the actual capacity of the battery under different aging states in order to determine whether the estimated battery capacity is accurate.

[0042] Optionally, the difference between the actual capacity and the estimated capacity of the battery under different aging conditions can be calculated, and the difference can be compared with a reference error value. If the difference is less than the parameter error value, it means that the estimated battery capacity is relatively accurate. If the difference is not less than the reference error value, it means that the estimated battery capacity is inaccurate. The parameters of the battery capacity estimation model can be adjusted based on the difference to improve the accuracy of the estimated battery capacity.

[0043] In steps S101 to S104, the current and voltage information of batteries under different aging states can be obtained first. Then, the health indicators of batteries under different aging states are determined based on the capacity estimation model. Based on the correlation function between the health indicators and the actual capacity of the battery, the battery capacity under different aging states is estimated. Then, based on the actual capacity of the battery under different aging states, error analysis is performed on the estimated battery capacity under different aging states to determine whether the battery capacity estimation result is accurate. Since the correlation function is the correlation function between the battery health indicator parameters and the actual capacity, the battery capacity determined based on this correlation function is relatively accurate, achieving the purpose of improving the robustness of battery capacity estimation. Finally, error analysis can be performed on the estimated battery capacity to further improve the accuracy of battery capacity prediction, achieving the technical effect of improving the accuracy of battery capacity estimation and solving the technical problem of low accuracy of battery capacity prediction.

[0044] The method described in this embodiment will be further described below.

[0045] As an optional embodiment, step S102 involves inputting current data and voltage data into a capacity estimation model for processing to obtain battery health indicators under different aging states. This includes: performing spectral entropy analysis on current data and voltage data to obtain frequency domain features; performing information energy analysis on frequency domain features to obtain an energy information matrix; and performing singular value decomposition on the energy information matrix to obtain battery health indicators under different aging states.

[0046] In this embodiment, spectral entropy analysis can be performed on the current data and voltage data first, whereby the spectral entropy is used to characterize the frequency domain features corresponding to the current signal and voltage signal.

[0047] For example, a discrete random variable y = [y1, y2, ..., yn] can be predefined regarding current or voltage data, where n represents the variable length. If pi = p(yi) represents the probability of yi occurring, then... Then its information entropy can be given by the following formula:

[0048]

[0049] Optionally, after determining the information entropy, the spectral entropy can be determined based on the information entropy, where the spectral information entropy can be represented by SEN, and the spectral entropy SEN can be determined by the following formula:

[0050]

[0051] Where Yi is the i-th power spectrum of the original signal y, and N is the total number of power spectra. To make SEN values ​​comparable, SEN can be standardized according to the following formula:

[0052]

[0053] Optionally, after determining the spectral entropy, frequency domain features can be further obtained based on the spectral entropy. After obtaining the frequency domain features, a detailed energy analysis can be performed on the frequency domain features to obtain the information energy matrix.

[0054] For example, the information energy at time slice (m-1) can be determined using the following formula:

[0055]

[0056] Here, H indicates the spectral information entropy (SEN). From the above formula, it can be seen that the information energy at the (m-1)th time slice is the sum of all information entropies up to the (m-1)th instant. Therefore, this definition can introduce rich information because most of the information in E(m) is already contained in E(m-1). The instantaneous information energy can be improved using the following formula to eliminate information redundancy:

[0057]

[0058] By combining all the instantaneous energy information, a vector describing the energy of the process information can be obtained, namely... ,in This is the instantaneous total. It extends the information vector of a specific sensor to multiple sensors, and assigns each sensor's... Arranging them in different columns yields an information energy matrix. As shown in the following formula:

[0059]

[0060] in, The total number of sensors, Indicates the first The first sensor The information energy of a time slice. From The structure of the matrix shows that... It is a combination of information energy from multiple sensors at multiple moments.

[0061] Optionally, after obtaining the information energy matrix, singular value decomposition can be performed on the energy information matrix to obtain the health indicators of the battery under different aging states.

[0062] As an optional implementation, singular value decomposition is performed on the energy information matrix to obtain the battery health index under the different aging states, including: performing singular value decomposition on the energy information matrix to obtain at least one sub-matrix and the singular values ​​corresponding to each sub-matrix; using the singular values ​​corresponding to each sub-matrix to extract signal features in the energy information matrix; and determining the battery health index based on the signal features in the energy information matrix.

[0063] In this embodiment, it is assumed that the energy information matrix Its rank is . eigenvalues express, Called a matrix The singular values. Decomposing based on singular values ​​will result in two matrices. and And has a diagonal matrix satisfy: Since D is a diagonal matrix, singular value decomposition is equivalent to transforming a matrix into a matrix with rank D. The matrix F is decomposed into a weighted sum of matrices of rank 1 and dimension m×n, as shown in the following formula:

[0064]

[0065] Specifically, singular value decomposition of matrix F yields a series of submatrices Fi and their corresponding singular values. .

[0066] Optionally, matrix singular values ​​have scale invariance and rotation invariance, which meet the requirements of stability, rotation invariance and scale invariance for feature extraction in pattern recognition. Therefore, matrix singular values ​​can be used to extract signal features in pattern recognition.

[0067] Optionally, after obtaining the signal features, the battery health indicators can be further extracted from the signal features, and then the estimated capacity of the battery can be obtained based on the battery health indicators.

[0068] As an optional implementation, before performing step S102, the method further includes: extracting features from the current data samples and voltage data samples of the battery samples; determining the health indicators of the battery samples based on the extracted features; performing linear regression between the health indicators of the battery samples and the calibration capacity to obtain the regression results; and establishing an initial capacity estimation model based on the current data samples, voltage data samples, and regression results.

[0069] In this embodiment, experiments can be designed to obtain current and voltage data of battery samples under different aging conditions. The obtained current and voltage data can then be used as current and voltage data templates for battery samples. Subsequently, based on the characteristic features of the current and voltage data samples, health indicators of batteries under different aging conditions can be obtained. Then, the health indicators can be linearly regressed with the standard capacity of the battery to obtain a capacity estimation model.

[0070] As an optional implementation, step S104 involves performing error analysis on the estimated capacity of the battery based on the actual capacity to obtain an estimated battery capacity result, including: obtaining the relative error and average error between the actual capacity and the estimated capacity; and determining the estimation result in response to the relative error being less than or equal to a relative error threshold and the average error being less than or equal to an average error threshold, wherein the relative error threshold is a preset condition for evaluating whether the relative error is reasonable and the average error threshold is a preset condition for evaluating whether the average error is reasonable.

[0071] In this embodiment, the estimated battery capacity under different aging conditions can be estimated based on the actual capacity of the battery under different aging conditions obtained in advance. For example, the relative error and average error between the actual capacity and the estimated capacity of the battery under different aging conditions can be calculated, and the calculated relative error can be compared with a relative error threshold, and the calculated average error can be compared with an average error threshold to determine whether the estimated battery capacity under different aging conditions is accurate.

[0072] Optionally, if the calculated relative error is less than or equal to the relative error threshold, it indicates that the estimated battery capacity is relatively accurate; if the calculated relative error is greater than the relative error threshold, it indicates that the estimated battery capacity is inaccurate. Similarly, if the calculated average error is less than or equal to the average error threshold, it indicates that the estimated battery capacity is relatively accurate; if the calculated average error is greater than the average error threshold, it indicates that the estimated battery capacity is inaccurate.

[0073] As an optional implementation, in response to a relative error greater than the relative error threshold and / or an average error greater than the average error threshold, the initial capacity estimation model is subjected to multiple linear regressions using the relative error and the average error until the initial capacity estimation model converges; the converged initial capacity estimation model is then determined as the capacity estimation model.

[0074] In this embodiment, in response to a relative error greater than a relative error threshold, multiple linear regression calculations can be performed on the initial capacity estimation model based on the relative error until the initial capacity estimation model converges. Similarly, in response to an average error greater than an average error threshold, multiple linear regression calculations can be performed on the initial capacity estimation model based on the average error until the initial capacity estimation model converges. The converged initial capacity regression model is then determined as the capacity estimation model.

[0075] Example 2

[0076] The technical solutions of the embodiments of the present invention will be illustrated below with reference to preferred embodiments.

[0077] To address the increasingly prominent contradiction between fuel supply and demand and environmental pollution, electric vehicles (EVs) have received widespread attention both domestically and internationally due to their non-reliance on fossil fuels and zero emissions during operation. As the power source of EVs, the power battery plays a crucial role in their performance. Power batteries can be categorized based on electrode materials into lead-acid batteries, nickel-cadmium batteries, nickel-metal hydride batteries, and lithium-ion batteries. Among these, lithium-ion batteries dominate the EV industry due to their high energy density, long cycle life, wide temperature adaptability, and low self-discharge rate. To extend battery life and ensure the reliability of EVs, accurately estimating the battery's state-of-health (SOH) is essential. Currently, battery SOH is primarily defined as the ratio of actual usable capacity to initial usable capacity. Generally, the initial capacity is a known fixed value; therefore, estimating the battery SOH is equivalent to estimating the current actual usable capacity of the power battery. Because power batteries gradually age during use, some active lithium ions are lost due to irreversible chemical reactions, thus the battery SOH generally declines with battery use. Because battery SoH (Power Output) cannot be directly measured by sensors such as voltage and current, and because the usage scenarios of electric vehicles are highly dynamic and complex, it is difficult to accurately estimate battery SoH. Inaccurate SoH estimation not only compromises the safety and reliability of the power battery, but also, due to the coupling relationship between SoH and the remaining battery capacity (State of Charge, or SOC), severely affects the accuracy of SOC estimation.

[0078] Currently, battery SoH estimation methods can be divided into three categories: (1) electrochemical methods; (2) equivalent circuit methods; and (3) data-driven methods. Among them, the electrochemical method is mainly based on the aging mechanism of the battery itself and has the best estimation accuracy among the three methods. However, the model involves too many parameters, and the calibration and updating of it are cumbersome, making it difficult to apply in battery management systems (BMS) with limited computing resources. To this end, the equivalent circuit method with fewer model parameters has been proposed. This method is popular because of its good balance between computational burden and model accuracy. However, it relies on online parameter identification algorithms such as recursive least squares algorithm to update model parameters, and the robustness of the model still faces great challenges. In addition, this method is usually combined with filtering algorithms such as particle filtering or Kalman filtering. If the prior probability distribution or noise variance is not properly predefined in these filtering algorithms, the divergence of the method cannot be guaranteed.

[0079] To overcome the shortcomings of electrochemical and equivalent circuit methods, a data-driven approach is proposed. This method, by mining large amounts of data, can directly analyze the intrinsic relationship between the battery's SoH (Solar Energy Flow) and its related influencing factors (commonly referred to as health indicators). Its key advantage is that it does not require prior in-depth understanding of the complex battery aging mechanism. Furthermore, since the model is calibrated during training, its computational burden during application is relatively light. The selection of health indicators has a significant impact on the model's estimation performance. To date, many health indicators have been extracted from battery current, voltage, etc. Although these health indicators have a certain correlation with battery degradation, their relationship is non-linear. Non-linear models such as Support Vector Machines (SVM) or Grey Relational Analysis (GRA) are used to map health indicators to battery capacity. However, due to the non-linear characteristics, they are to some extent uninterpretable and unstable. Therefore, when the non-linear model is incorrectly calibrated or features are incorrectly extracted or calculated, the robustness of the estimated SoH may decrease. Thus, all three methods for estimating the SoH of batteries mentioned above suffer from the technical problem of low estimation accuracy.

[0080] However, embodiments of the present invention propose using information energy theory and singular value decomposition to estimate battery capacity under different aging conditions. The battery capacity estimation method provided by the embodiments of the present invention will be further described below:

[0081] Figure 2 This is a schematic diagram of a battery capacity estimation method provided according to an embodiment of the present invention. Figure 2As shown, estimating the battery capacity under different aging conditions can be achieved through two main parts. The first part is mainly used for health indicator extraction, and the second part is mainly used for battery capacity estimation. The health indicator extraction mainly includes steps such as data extraction, singular value decomposition, spectral entropy analysis, and information energy analysis. The battery capacity estimation is mainly based on the extracted health indicators to predict the battery capacity.

[0082] Figure 3 This is a flowchart of another battery capacity estimation method according to an embodiment of the present invention, such as... Figure 3 As shown, the method may include the following steps:

[0083] Step S301, data extraction.

[0084] In this embodiment, current and voltage data of lithium-ion batteries can be obtained from a large number of experiments.

[0085] Step S302, Spectral entropy analysis.

[0086] In this embodiment, spectral entropy analysis can be performed on the current and voltage data. A discrete random variable y = [y1, y2, ..., yn] is given in advance regarding the current or voltage data, where n represents the variable length. If pi = p(yi) represents the probability of yi occurring, then... Then its information entropy can be given by the following formula:

[0087]

[0088] According to the definition of information entropy, H(y) is only related to the probability distribution of variable y and is not related to the specific value of variable y. This means that information entropy can effectively avoid the interference of noise to a certain extent.

[0089] Spectral information entropy, often simply referred to as spectral entropy, is one of the entropy concepts that has been successfully applied in fields such as speech recognition, mechanical fault analysis, and medical treatment. This invention uses the spectral information entropy method to extract the frequency domain features of a signal, where spectral information entropy can be represented by SEN. Let Y be the Fourier transform of signal y, then SEN can be defined as:

[0090]

[0091] Where Yi is the i-th power spectrum of the original signal y, and N is the total number of power spectra. To make SEN values ​​comparable, SEN can be standardized according to the following formula:

[0092]

[0093] As can be seen from the above formula, SEN calculates the uncertainty of a signal in the frequency domain, not the time domain. Therefore, SEN can evaluate the spectral structure of a signal.

[0094] Step S303, Information Energy Analysis.

[0095] In this embodiment, information energy is a concept derived from the term "energy" in thermodynamics. It is a generalization of energy in information theory. The most important characteristic of information energy is that it integrates information from multiple times and multiple sensors, greatly enhancing its robustness to uncertainties such as sensor noise and sudden failures. For a specific sensor, the information energy at time slice (m-1) can be expressed by the following formula:

[0096]

[0097] Here, H indicates the spectral information entropy (SEN). From the above formula, it can be seen that the information energy at the (m-1)th time slice is the sum of all information entropies up to the (m-1)th instant. Therefore, this definition can introduce rich information because most of the information in E(m) is already contained in E(m-1). The instantaneous information energy can be improved using the following formula to eliminate information redundancy:

[0098]

[0099] By combining all the instantaneous energy information, a vector describing the energy of the process information can be obtained, namely... ,in This is the instantaneous total. It extends the information vector of a specific sensor to multiple sensors, and assigns each sensor's... Arranging them in different columns yields an information energy matrix. As shown in the following formula:

[0100]

[0101] in, The total number of sensors, Indicates the first The first sensor The information energy of a time slice. From The structure of the matrix shows that... It is a combination of information energy from multiple sensors at multiple instants. If a sensor suddenly malfunctions within a short period of time, or if a sensor value becomes abnormal at a certain moment, from... The final features extracted may still be reliable because information from other sensors or other moments can, to some extent, correct for the aforementioned uncertainties.

[0102] This invention uses only current and voltage data for feature extraction. Therefore, the information energy matrix... There are only two columns. However, this method provides a general framework for incorporating more sensor data into the information energy matrix. This further improves the robustness of subsequent feature extraction.

[0103] Step S304, singular value decomposition.

[0104] In this embodiment, due to the derived energy information matrix Its size is too large to be directly used for estimating the SoH of the battery. Therefore, it is necessary to use a different approach. This involves extracting more abstract features for practical applications. Existing feature extraction methods include basic operations such as calculating the mean or standard deviation of each column. However, these methods ignore the correlation between different columns, i.e., they ignore information sharing between different sensors. Therefore, an effective method for extracting more useful features should be to extract more abstract features from the matrix. Considering it as a whole, this invention uses Singular Value Decomposition (SVD) to obtain the final features, also known as health indicators. Hypothesis Matrix Its rank is . eigenvalues In other words, then Called a matrix The singular values. According to singular value decomposition, there must exist two matrices. and And has a diagonal matrix satisfy: Since D is a diagonal matrix, SVD is equivalent to transforming a matrix of rank D into a matrix of rank D. The matrix F is decomposed into a weighted sum of matrices of rank 1 and dimension m×n, as shown in the following formula:

[0105]

[0106] Specifically, singular value decomposition of matrix F yields a series of submatrices Fi and their corresponding singular values. Singular values ​​reflect the amount of time-frequency information contained in the matrix and, to a certain extent, represent the inherent characteristic modes of the matrix. Furthermore, matrix singular values ​​possess scale invariance and rotation invariance, satisfying the stability, rotation invariance, and scale invariance requirements for feature extraction in pattern recognition. Therefore, matrix singular values ​​are frequently used to extract signal features in pattern recognition.

[0107] Considering the above advantages, this invention employs singular value decomposition (SVD) to extract health indicators from the information energy matrix F. Since this paper constructs F using only current and voltage data, F has two columns. Therefore, two health indicators can be extracted from F to represent the battery's aging characteristics.

[0108] Step S305, battery capacity estimation.

[0109] In this embodiment, the dataset is divided into a training dataset and a test dataset. The data in the training dataset is used to train the model, and the data in the test dataset is used to test the model. By extracting features from the current and voltage data in the training dataset, health indicators under different aging states can be obtained. Then, a linear regression is performed between the health indicators and the calibration capacity to obtain a capacity estimation model. Finally, the health indicators of the test dataset are calculated based on the linear capacity estimation model to obtain the estimated capacity. By analyzing the error between the estimated capacity and the reference capacity, the accuracy and robustness of the model are verified.

[0110] Figure 4 This is a schematic diagram illustrating the change of a first health indicator with battery capacity according to an embodiment of the present invention, as shown below. Figure 4 As shown, the first health indicator decreases as battery capacity increases.

[0111] Figure 5 This is a schematic diagram illustrating the change in battery capacity according to a second health indicator based on an embodiment of the present invention, as shown below. Figure 5 As shown, the second health indicator decreases with increasing battery capacity. Based on Figure 4 and Figure 5 It can be seen that the health indicators decrease as the battery capacity increases. The higher the battery capacity, the higher the battery health indicators, and the lower the battery capacity, the lower the battery health indicators.

[0112] Figure 6 This is a schematic diagram illustrating the estimated capacity and actual capacity according to an embodiment of the capacity estimation method of the present invention, as shown below. Figure 6 As shown, the estimated capacity of the battery is close to the battery capacity estimated by the capacity estimation method, indicating that the capacity estimation method has high accuracy.

[0113] Figure 7 This is a schematic diagram illustrating the relative error and average error between the estimated capacity and the actual capacity according to an embodiment of the present invention. Figure 7 As shown, the relative error of the battery is close to the actual error by 1.75%.

[0114] Example 3

[0115] According to an embodiment of the present invention, a battery capacity determination device is also provided. It should be noted that this battery capacity determination device can be used to execute the battery capacity determination method in Embodiment 1.

[0116] Figure 8 This is a schematic diagram of a battery capacity determination device according to an embodiment of the present invention. Figure 8 As shown, the battery capacity determination device 800 may include: an acquisition unit 801, a processing unit 802, a determination unit 803, and an analysis unit 804.

[0117] Acquisition unit 801 is used to acquire current data and voltage data in the battery;

[0118] The processing unit 802 is used to input current data and voltage data into the capacity estimation model for processing to obtain the health indicators of the battery under different aging states. The capacity estimation model is used to determine the characteristics of the battery under different aging states.

[0119] The determining unit 803 is used to determine the estimated capacity of the battery based on the health indicators of the battery under different aging states and the correlation function between the health indicators and the actual capacity of the battery, wherein the correlation function is used to characterize the linear relationship between the health indicators and the actual capacity.

[0120] Analysis unit 804 is used to perform error analysis on the estimated capacity of the battery based on the actual capacity, and to determine the estimated battery capacity result.

[0121] Optionally, the processing unit 802 includes: a first analysis module for performing spectral entropy analysis on current and voltage data to obtain frequency domain features; a second analysis module for performing information energy analysis on the frequency domain features to obtain an energy information matrix; and a decomposition module for performing singular value decomposition on the energy information matrix to obtain health indicators of the battery under different aging states.

[0122] Optionally, the decomposition module is further configured to: perform singular value decomposition on the energy information matrix to obtain at least one submatrix and the singular values ​​corresponding to each submatrix; extract signal features from the energy information matrix using the singular values ​​corresponding to each submatrix; and determine the health indicators of the battery based on the signal features in the energy information matrix.

[0123] Optionally, the method further includes: a feature extraction unit for extracting features from current data samples and voltage data samples of the battery sample; a first determination unit for determining the health indicators of the battery sample based on the extracted features; a first linear regression unit for performing linear regression between the health indicators of the battery sample and the calibration capacity to obtain regression results; and a generation unit for establishing an initial capacity estimation model based on the current data samples, voltage data samples, and regression results.

[0124] Optionally, the analysis unit 804 includes: a first acquisition unit, used to acquire the relative error and average error between the actual capacity and the estimated capacity; and a second determination unit, used to determine the estimation result when the relative error is less than or equal to a relative error threshold and the average error is less than or equal to an average error threshold, wherein the relative error threshold is a preset condition for evaluating whether the relative error is reasonable and the average error threshold is a preset condition for evaluating whether the average error is reasonable.

[0125] Optionally, the method further includes: a second linear regression unit, used to perform multiple linear regressions on the initial capacity estimation model using the relative error and the average error in response to the relative error being greater than a relative error threshold and / or the average error being greater than an average error threshold, until the initial capacity estimation model converges; and a third determination unit, used to determine the converged initial capacity estimation model as the capacity estimation model.

[0126] In this embodiment, an acquisition unit is used to acquire current and voltage data from the battery; a processing unit is used to input the current and voltage data into a capacity estimation model for processing to obtain the battery's health indicators under different aging states, wherein the capacity estimation model is used to determine the battery's characteristics under different aging states; a determination unit is used to determine the estimated capacity of the battery based on the battery's health indicators under different aging states and the correlation function between the health indicators and the battery's actual capacity, wherein the correlation function is used to characterize the linear relationship between the health indicators and the actual capacity; and an analysis unit is used to perform error analysis on the estimated capacity of the battery based on the actual capacity to determine the estimated battery capacity result. In other words, in this embodiment of the invention, the battery's health indicators under different aging states can be obtained first based on the capacity estimation model. Then, the estimated capacity of the battery can be determined based on the correlation function between the health indicators and the battery's actual capacity. Since the correlation function is the correlation function between the battery's health indicator parameters and the actual capacity, the battery capacity determined based on this correlation function is relatively accurate, achieving the goal of improving the robustness of battery capacity estimation. Finally, error analysis can be performed on the estimated capacity of the battery to further improve the accuracy of battery capacity prediction, achieving the technical effect of improving the accuracy of battery capacity estimation and solving the technical problem of low accuracy in battery capacity prediction.

[0127] Example 4

[0128] According to an embodiment of the present invention, a computer-readable storage medium is also provided, the storage medium including a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to execute the battery capacity determination method in Embodiment 1.

[0129] Example 5

[0130] According to an embodiment of the present invention, a processor is also provided for running a program, wherein the program executes the battery capacity determination method in Embodiment 1 during runtime.

[0131] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0132] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0133] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of modules can be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection between units or modules, and can be electrical or other forms.

[0134] The units described as separate components may or may not be physically separate. Similarly, the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0135] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0136] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0137] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for determining battery capacity, characterized in that, include: Obtain battery current and voltage data; The current data and voltage data are input into the capacity estimation model for processing to obtain the health indicators of the battery under different aging conditions. The capacity estimation model is used to determine the performance data of different batteries under different aging conditions. Based on the health indicators of the battery under different aging conditions, and the correlation function between the health indicators and the actual capacity of the battery, the estimated capacity of the battery is determined, wherein the correlation function is used to characterize the linear relationship between the health indicators and the actual capacity. An error analysis is performed on the estimated capacity of the battery based on the actual capacity to determine the estimated capacity of the battery.

2. The method according to claim 1, characterized in that, The current data and voltage data are input into a capacity estimation model for processing to obtain the health indicators of the battery under different aging states, including: Spectral entropy analysis is performed on the current data and the voltage data to obtain frequency domain characteristics; Information energy analysis is performed on the frequency domain features to obtain an energy information matrix; Singular value decomposition is performed on the energy information matrix to obtain the health indicators of the battery under different aging states.

3. The method according to claim 2, characterized in that, Singular value decomposition is performed on the energy information matrix to obtain the battery's health indicators under different aging states, including: Singular value decomposition is performed on the energy information matrix to obtain at least one submatrix and the singular values ​​corresponding to each submatrix; Signal features in the energy information matrix are extracted using the singular values ​​corresponding to each of the sub-matrices; The health indicators of the battery are determined based on the signal characteristics in the energy information matrix.

4. The method according to claim 1, characterized in that, Before inputting the current data and voltage data of the battery into the capacity estimation model for processing to obtain the health indicators of the battery under different aging states, the method further includes: Feature extraction is performed on the current and voltage data samples of the battery samples; The health indicators of the battery samples are determined based on the extracted features; Linear regression was performed on the health indicators and calibration capacity of the battery samples to obtain the regression results; An initial capacity estimation model is established based on the current data sample, the voltage data sample, and the regression results.

5. The method according to claim 1, characterized in that, The step of performing error analysis on the estimated capacity of the battery based on the actual capacity to obtain the estimated battery capacity includes: Obtain the relative error and average error between the actual capacity and the estimated capacity; In response to the relative error being less than or equal to a relative error threshold and the average error being less than or equal to an average error threshold, the estimation result is determined, wherein the relative error threshold is a preset condition for evaluating whether the relative error is reasonable, and the average error threshold is a preset condition for evaluating whether the average error is reasonable.

6. The method according to claim 5, characterized in that, The method further includes: In response to the relative error being greater than the relative error threshold, and / or the average error being greater than the average error threshold, the initial capacity estimation model is subjected to multiple linear regressions using the relative error and the average error until the initial capacity estimation model converges; The converged initial capacity estimation model is determined as the capacity estimation model.

7. A device for determining battery capacity, characterized in that, include: The acquisition unit is used to acquire current and voltage data from the battery. The processing unit is used to input the current data and voltage data into the capacity estimation model for processing to obtain the health indicators of the battery under different aging states, wherein the capacity estimation model is used to determine the characteristics of the battery under different aging states; A determining unit is used to determine the estimated capacity of the battery based on the health indicators of the battery under different aging states and the correlation function between the health indicators and the actual capacity of the battery, wherein the correlation function is used to characterize the linear relationship between the health indicators and the actual capacity. An analysis unit is used to perform error analysis on the estimated capacity of the battery based on the actual capacity, and to determine the estimated capacity of the battery.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored program, wherein, when the program is executed, it controls the device on which the computer-readable storage medium is located to perform the method according to any one of claims 1 to 6.

9. A processor, characterized in that, The processor is used to run a program, wherein the program is executed by the processor to perform the method according to any one of claims 1 to 6.

10. A vehicle, characterized in that, The vehicle is used to perform the method according to any one of claims 1 to 6.

Images