A method for estimating battery capacity, an electronic device, and a readable storage medium

Abstract

This application belongs to the technical field of battery state assessment, and particularly relates to a method for estimating battery capacity, an electronic device, and a readable storage medium. This method can solve the problem of low accuracy in estimating battery capacity in the prior art. The method includes obtaining charging data of the battery to be measured within a target voltage segment, where the sum of the correlations between multiple health indicators corresponding to the charging data and the battery capacity is the highest among all voltage segments; determining multiple health indicators of the charging data; inputting the multiple health indicators into a trained neural network model to obtain the battery capacity of the battery to be measured, and the neural network model is obtained by training with training samples, and the training samples are used to represent the corresponding relationship between the battery capacity and the multiple health indicators.

Description

Technical Field

[0001] This application belongs to the technical field of battery state assessment, and particularly relates to a method for estimating battery capacity, an electronic device, and a readable storage medium. Background Art

[0002] Battery capacity is an important indicator for determining the health status and remaining service life of a battery, which directly affects the reliability and safety of the battery. Therefore, it is particularly important to accurately estimate the battery capacity.

[0003] Currently, a battery capacity estimation model is usually used to estimate the capacity of a battery under test. The battery capacity estimation model is trained using charge and discharge data obtained under experimental conditions where a sample battery is fully charged and discharged in a cycle (i.e., the process where the state of charge (SOC) of the battery discharges from 100% to 0%, and then charges from 0% to 100%). However, in actual use, the battery generally will not be used until the power is completely exhausted, that is, it is charged when the SOC is 0%. In most cases, the battery starts to be charged when the SOC of the battery is between 10% and 20%, and the charging ends when the SOC of the battery is between 80% and 100%. The entire charging process has strong randomness. Therefore, for the charge and discharge data in the actual use process, the accuracy of the battery capacity estimated by using this kind of battery capacity estimation model is relatively low. Summary of the Invention

[0004] In view of this, embodiments of this application provide a method for estimating battery capacity, an electronic device, and a readable storage medium to solve the problem that the accuracy of estimating battery capacity in the prior art is relatively low.

[0005] The first aspect of the embodiments of this application provides a method for estimating battery capacity, and the method includes: obtaining the charging data of the battery under test within a target voltage segment, where the sum of the correlations between the multiple health indicators corresponding to the charging data and the battery capacity is the highest among all voltage segments; determining the multiple health indicators of the charging data; inputting the multiple health indicators into a trained neural network model to obtain the battery capacity of the battery under test, and the neural network model is trained by using training samples, and the training samples are used to represent the corresponding relationship between the battery capacity and the multiple health indicators.

[0006] In combination with the first aspect, in the first possible implementation manner of the first aspect, the multiple health indicators include M data domain health indicators and N model domain health indicators, M + N ≥ 2, M ≥ 0, N ≥ 0; the data domain health indicators include at least one of segment capacity, peak value of the incremental capacity analysis (ICA) curve, voltage corresponding to the peak value of the ICA curve, charging time, and voltage rise per unit time; the model domain health indicators include at least one of ohmic internal resistance, polarization internal resistance, polarization capacitance, and time constant.

[0007] In combination with the first aspect, in the second possible implementation manner of the first aspect, the training samples include original battery samples and virtual battery samples. The virtual battery samples are generated based on the original battery samples, and the original batteries corresponding to the original battery samples are of the same type as the battery to be measured.

[0008] In combination with the first aspect, in the third possible implementation manner of the first aspect, the virtual battery samples are generated in the following manner: Obtain the actual aging data of the original battery. The actual aging data includes the change data of time, voltage, and current corresponding to each cycle charge and discharge process during the entire process in which the battery capacity of the original battery decays from 100% to 0%. Determine the target voltage segment according to the actual aging data and multiple health indicator types. Generate the original battery sample according to the charge data within the target voltage segment. Expand the battery capacity and each health indicator in the original battery sample to generate virtual battery samples.

[0009] In combination with the first aspect, in the fourth possible implementation manner of the first aspect, determining the target voltage segment according to the actual aging data and multiple health indicator types includes: Determine multiple voltage segments according to the actual aging data. Among them, in each cycle charge and discharge process, each voltage segment has corresponding voltage segment data. Screen the voltage segment data according to the constraint conditions R is the voltage segment width, and the starting SOC is less than or equal to the ending SOC. Determine the F value of the screened voltage segment data within each voltage segment, where, r i is the score of the correlation between the i-th health indicator of the voltage segment data and the battery capacity. Determine the voltage segment corresponding to the maximum F value as the target voltage segment.

[0010] In combination with the first aspect, in the fifth possible implementation manner of the first aspect, expanding the battery capacity and each health indicator in the original battery sample to generate virtual battery samples includes: Determine the characteristic data center, left skewness, right skewness, and variance of the battery capacity and each health indicator in the original battery sample. Determine the upper expansion bound and lower expansion bound of the battery capacity and each health indicator according to the characteristic data center, left skewness, right skewness, and variance. Expand the battery capacity and each health indicator according to the upper expansion bound and lower expansion bound of the battery capacity and each health indicator to generate virtual battery samples that meet the preset conditions.

[0011] In combination with the first aspect, in the sixth possible implementation manner of the first aspect, the neural network model is a long short-term memory (LSTM) neural network model.

[0012] In combination with the first aspect, in the seventh possible implementation manner of the first aspect, the LSTM neural network model includes an LSTM neural network with 2 layers, 128 hidden nodes, and a time step of 5.

[0013] A second aspect of the embodiments of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the method according to any one of the first aspect are implemented.

[0014] A third aspect of the embodiments of the present application provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps of the method according to any one of the first aspect are implemented.

[0015] The beneficial effects of the embodiments of the present application compared with the prior art are as follows: The present embodiment provides a method, device, and readable storage medium for estimating battery capacity. The method processes the charging data of the battery to be measured within the target voltage segment through a trained neural network model to obtain the battery capacity of the battery to be measured. The charging data used for estimation in this embodiment is the charging data with the highest sum of the correlations between multiple health indicators and the battery capacity in each voltage segment. Therefore, the battery capacity of the battery to be measured estimated through this charging data is relatively accurate, and the data required in the estimation method is small, which can improve the estimation efficiency of the battery capacity. At the same time, the neural network model provided in this embodiment is trained through the battery capacity and multiple health indicators, rather than a fixed single health indicator. Therefore, the accuracy of the model is higher. Description of the Drawings

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0017] Figure 1 It is a flowchart of a method for constructing a neural network model provided by an embodiment of the present application;

[0018] Figure 2 It is a schematic diagram of the first-order RC equivalent circuit model of the original battery provided by an embodiment of the present application;

[0019] Figure 3 It is a schematic diagram of the extended data upper and lower bound ranges of the MD-MTD algorithm provided by an embodiment of the present application;

[0020] Figure 4It is a schematic diagram of the virtual battery sample generation process provided by an embodiment of the present application;

[0021] Figure 5 It is a schematic diagram of the generation result of virtual aging data of the capacity attenuation curve provided by an embodiment of the present application;

[0022] Figure 6 It is a schematic diagram of generating virtual battery samples from original batteries provided by an embodiment of the present application;

[0023] Figure 7 It is a schematic diagram of the process of the battery capacity estimation method provided by an embodiment of the present application;

[0024] Figure 8 It is a schematic diagram of the estimation result of the CS2-36 battery capacity using the trained LSTM neural network model provided by an embodiment of the present application;

[0025] Figure 9 It is a schematic diagram of the estimation result of the CS2-37 battery capacity using the trained LSTM neural network model provided by an embodiment of the present application;

[0026] Figure 10 It is a schematic diagram of an electronic device provided by an embodiment of the present application. Detailed implementation manners

[0027] In the following description, specific details such as specific system architectures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0028] The technical solutions provided by the present application are explained in detail below in conjunction with specific embodiments.

[0029] To solve the problem of low accuracy of estimation results in traditional battery capacity estimation methods, an embodiment of the present application provides a battery capacity estimation method. This method processes the charging data of a battery under test within a target voltage segment through a trained neural network model to obtain the battery capacity of the battery under test. The method provided in this embodiment can improve the accuracy of battery capacity estimation results.

[0030] Figure 1 It is a schematic diagram of the process of the neural network model construction method provided by an embodiment of the present application. Refer to Figure 1 As shown, the method includes the following steps S101-S104.

[0031] S101. The electronic device obtains the actual aging data of the original battery, where the original battery and the battery to be tested are of the same type.

[0032] In this embodiment, batteries of the same type refer to batteries with the same battery model and working conditions. For example, the three batteries with model numbers CS2-35, CS2-36, and CS2-37 disclosed in the lithium battery aging dataset (Center for Advanced Life Cycle Engineering, CALCE) of the University of Maryland, USA are batteries of the same type.

[0033] In some embodiments, the actual aging data of the original battery is obtained through a complete aging cycle experiment on the original battery. The complete aging cycle experiment of the original battery is a process of performing cyclic charge and discharge on the original battery (for example, cycling 10,000 times) to decay the state of health (SOH) of the original battery from 100% to 0%, and recording the time and the voltage and current during the SOH decay process. Based on this, the actual aging data includes the change data of time, voltage, and current corresponding to each cyclic charge and discharge process during the entire process of the SOH of the original battery decaying from 100% to 0%.

[0034] S102. The electronic device determines the health indicators corresponding to each voltage segment during each cyclic charge and discharge process in the actual aging data.

[0035] Health factors (HF) refer to the index factors that affect the decay of battery capacity. For example, the amount of charge input by the battery within a certain SOC range (such as 20%-80%); or the charging time required for the battery to complete charging a certain SOC range.

[0036] In order to fully reflect the decay law of battery capacity and enhance the reliability of the constructed model, in this embodiment, multiple HFs with a relatively high correlation with battery capacity are extracted from the data domain and the model domain respectively. In this embodiment, the multiple HFs include M data domain HFs and N model domain HFs, where M + N ≥ 2, M ≥ 0, and N ≥ 0. Specifically, the data domain HFs include at least one of segment capacity, peak value of the incremental capacity analysis (ICA) curve, voltage corresponding to the peak value of the ICA curve, charging time, and voltage rise per unit time; the model domain HFs include at least one of ohmic internal resistance, polarization internal resistance, polarization capacitance, and time constant. The specific definitions of each HF are shown in Table 1 below.

[0037] Table 1 Health Indicator HF and Its Definition

[0038]

[0039] In an example of the present application, each HF in the data domain HF and each HF in the model domain HF are used as index factors to train the model, so as to improve the estimation accuracy of the trained model.

[0040] Based on the above, when the electronic device determines the values of the health indicators corresponding to each voltage region, first, according to the actual aging data, a plurality of voltage segments are determined; wherein, in each charge-discharge cycle, each voltage segment has corresponding voltage segment data. The voltage segment data corresponding to each voltage segment includes the time, current, and voltage data within the voltage segment. Then, according to the voltage segment data, the health indicators corresponding to each voltage segment in different cycle charge-discharge processes are determined. In this embodiment, the health indicators corresponding to each voltage segment are obtained by calculating the time, current, and voltage data within the voltage segment. Among them, the data domain HF is derived from the relevant data generated during the constant current charging stage of the original battery; the model domain HF is the relevant data obtained through least squares method parameter identification based on the first-order RC equivalent circuit model of the original battery. See Figure 2 as shown in

[0041] It should be understood that the entire charging process of the battery is divided into two stages. The first stage is the constant current charging stage, that is, in this stage, the battery is charged with a constant charging current (usually 1C). For example, if the battery capacity of the battery is 2 ampere-hours (2Ah), then 1C means charging the battery with a current of 2A. During the constant current charging stage, the charging voltage of the battery changes and is monitored by sensors for monitoring the voltage and current during the battery charging process. When the sensor monitors that the battery is charged to the upper cut-off voltage (for example, for a single lithium-ion battery, the upper cut-off voltage is generally 4.2V), the battery starts to enter the second stage, that is, the constant voltage charging stage. In the second stage, the battery is charged with a constant voltage of 4.2V. In this stage, the charging current of the battery changes, and the sensor starts to monitor the charging current. When the charging current is less than 0.02 or 0.05, it is considered that the charging is completed. The data domain HF selected in this embodiment is determined according to the relevant data generated during the constant current stage (that is, the above-mentioned first stage) of the battery charging.

[0042] It should be noted that during the use of traditional batteries (such as lithium-ion batteries), the battery capacity will continuously decline over time. The selection of the influencing factor HF that characterizes the battery capacity attenuation is an important basis for battery capacity estimation and significantly affects the estimation effect. Therefore, in order to express the battery aging process as fully as possible, in this embodiment, HF is obtained from the data domain and the model domain respectively, and through the combined use of HF in the data domain and HF in the model domain, the composition of HF is enriched. Compared with the method of using the charge amount value charged within a fixed SOC interval (such as SOC between 20% and 80%) in the constant current charging stage as a single health indicator HF in the traditional battery capacity estimation model to achieve battery capacity estimation, the possibility of the estimation result being invalid when a single health indicator HF fails is reduced, and the reliability of the model is improved.

[0043] S103. The electronic device determines the optimal voltage segment according to the health indicators corresponding to each voltage segment in each cycle of charge and discharge.

[0044] The optimal voltage segment (which can also be referred to as the target voltage segment) refers to the voltage segment range with the highest correlation between each type of health indicator in HF and the battery capacity. It should be understood that in the method of using the peak value of the ICA curve within a fixed voltage segment as a single health indicator HF in the traditional battery capacity estimation model to achieve battery capacity estimation, generally, the peak value of the ICA curve will fall within a specific voltage segment (such as 3.9V - 4.1V), and this specific voltage segment is very likely to be inconsistent with the selected fixed voltage segment, or the selected fixed voltage does not include the peak value of the ICA curve, resulting in a low accuracy of the battery capacity estimation result based on this model. For each HF, there is a corresponding voltage segment range with the greatest correlation with the battery capacity. The optimal voltage segment proposed in this embodiment is the voltage segment range that can fully express the correlation between each HF and the battery capacity.

[0045] In order to better establish the mapping relationship between fragment data and battery capacity, it is necessary to improve the correlation between HF on fragment data and the full battery capacity as much as possible. Therefore, in this embodiment, an optimization objective function based on HF is constructed. At the same time, according to the actual situation of daily battery charging and use, constraint conditions are constructed, and the Quantum Particle Swarm Optimization (QPSO) algorithm is used to obtain the optimal voltage range. Usually, the charging process of the battery is random. For example, for a lithium-ion battery, generally, the battery is charged when the battery capacity decays to an SOC of about 5%-15%, and the charging stops when the SOC is about 80%-100%. Therefore, when selecting the optimal voltage fragment, not only the expression of the battery HF in this interval needs to be considered, but also the SOC change range during the actual use of the battery should be considered. In addition, the width of the selected voltage fragment is also a problem that needs to be considered. If the width of the voltage fragment is too wide, the possibility of covering the optimal interval during actual charging will decrease. At the same time, if the width of the voltage fragment is too narrow, the amount of data will become less, resulting in difficulty for HF to be fully expressed. Therefore, in this embodiment, when establishing the optimization objective function, not only the expression of HF is considered, but also the SOC change range during the actual use of the battery and the rationality of the voltage fragment width are comprehensively considered, so as to achieve accurate estimation of the battery capacity through fragment data. Based on the above conditions, the objective function and constraint conditions in this embodiment are shown as follows:

[0046]

[0047] In the formula, F is the sum of the correlations of each HF within a certain voltage fragment, that is, the total score of 9 HFs on a certain voltage fragment, r i is the score of the correlation between the i-th HF and the battery capacity, R is the width of the voltage fragment, and the starting SOC is less than or equal to the ending SOC. Among them, r i is calculated by the QPSO algorithm for the correlation between HF and the battery capacity.

[0048] In this embodiment, the electronic device randomly generates a series of voltage segments based on the actual aging data of the original battery through the QPSO algorithm. Each voltage segment includes corresponding time, current, and voltage data. Then, through the QPSO algorithm, within the randomly generated series of voltage segments, the target voltage segments that meet the constraint conditions in the above formula are screened based on the constraint conditions. Then, the F value of each target voltage segment in the target voltage segments is calculated, and the target voltage segment with the largest F value is used as the optimal voltage segment based on the objective function in the above formula. When the electronic device calculates the optimal voltage segment through the QPSO algorithm, it judges whether the change range of the SOC and the width of the voltage segment within each voltage segment meet the set constraint conditions. If they meet, the F value in the objective function is calculated; if not, the F value in the objective function is not calculated.

[0049] In one example, after selecting the optimal voltage segments for three batteries with model numbers CS2-35, CS2-36, and CS2-37 in CALCE through the above method, the optimal voltage segment for this type of battery is obtained as 3.91V - 4.13V.

[0050] During the actual use of the battery, in most cases, the battery starts to be charged when the SOC of the battery is between 10% and 20%, and ends until the SOC of the battery is between 80% and 100%. It can be seen that the charging process of the battery has a certain degree of randomness. Most traditional battery capacity estimation methods based on segment data use fixed voltage segments including the starting point or the ending point, without considering the expression of HF in different voltage segments of different batteries, which easily causes a decrease in estimation accuracy due to the inability to fully express HF. The method proposed in this embodiment comprehensively considers the expression of HF on the optimal voltage segment and the actual use process of the battery, and uses this as the objective function and constraint conditions, and uses the QPSO algorithm to screen the optimal voltage segment, so as to improve the expression ability of HF and improve the estimation accuracy.

[0051] S104. The electronic device generates an original battery sample according to the charging data within the optimal voltage segment.

[0052] The original battery sample is used to characterize the corresponding relationship between the battery capacity and multiple health indicators in the actual aging data of the original battery.

[0053] In this embodiment, the electronic device calculates the battery capacity of the original battery through the ampere-hour integration method according to the actual aging data, and extracts the health indicators corresponding to the charging data within the optimal voltage segment in the actual aging data.

[0054] S105. The electronic device performs data expansion on the battery capacity and each health indicator in the original battery sample to generate a virtual battery sample.

[0055] The virtual battery sample is used to characterize the correspondence between the battery capacity and multiple health indicators in the virtual aging data of the virtual battery.

[0056] In this embodiment, the electronic device, based on the battery capacity of the original battery and the HF extracted from the optimal voltage segment, uses the Multi-distribution mega-trend-diffusion (MD-MTD) algorithm to expand the data range, and adopts the QPSO optimization algorithm to generate random data within the expanded data range. Figure 3 It is a schematic diagram of the upper and lower bounds of the data expanded by the MD-MTD algorithm provided in this embodiment. The data expansion parameters and calculation methods are shown in Table 2 below.

[0057] Table 2 Data expansion parameters and calculation methods

[0058]

[0059] Figure 4 Schematic diagram of the virtual battery sample generation process provided in this embodiment. See Figure 4 As shown, the method includes the following steps: The electronic device determines the characteristic data center, left skewness, right skewness, and variance of the battery capacity and each health indicator in the original battery sample (S401). The electronic device calculates the expanded upper bound and expanded lower bound of the battery capacity and each health indicator using the MD-MTD algorithm according to the characteristic data center, left skewness, right skewness, and variance (S402). Furthermore, the electronic device expands the data of the battery capacity and each health indicator using the QPSO optimization algorithm according to the expanded upper bound and expanded lower bound of the battery capacity and each health indicator, and generates multiple groups of virtual aging data that meet the preset conditions (S403), where the preset conditions include: the mean absolute error (MAPE) between the generated virtual aging data and the battery capacity and health indicator data of the original battery is less than 0.01. Finally, the electronic device combines the virtual aging data of the battery capacity and the corresponding individual HFs in multiple groups of virtual aging data into a virtual battery sample, and generates several groups of virtual battery samples (S404).

[0060] Exemplarily, taking Figure 5 the generation process of the virtual aging data of the capacity decay curve as an example, what is obtained through the above virtual battery sample generation process is a virtual curve of capacity decay. In the same way, virtual curves of each of the 9 HFs can be generated, and then the virtual curves of each of the 9 HFs are combined one by one to generate a virtual battery sample.

[0061] To improve the generalization ability of the trained model as much as possible and at the same time take into account the computational overhead of the electronic device, the same method as above is used to obtain a preset number of virtual battery samples (for example, the preset number is 50). See Figure 6 As shown, it is a schematic diagram of generating virtual battery samples from the original battery data in this embodiment.

[0062] Since the battery capacity attenuation experiment takes a long time and has high environmental requirements, the number of battery samples available for the experiment is small, resulting in a weak generalization ability of the estimation model. In this embodiment, to solve the problem of the small number of battery samples, this patent uses the multi-distribution overall trend diffusion technology (MD-MTD) to expand the upper and lower bounds of the data, and uses the QPSO algorithm to find the virtual sample data that meets the requirements, thereby increasing the number of samples available for model training and enhancing the generalization ability of the model.

[0063] S106. The electronic device trains the neural network model based on the training samples, where the training samples include the original battery samples and the virtual battery samples.

[0064] In this embodiment, the neural network model can be a Long Short-Term Memory (LSTM) neural network model. Among them, the LSTM neural network model can obtain different basic architectures based on different numbers of network layers, hidden nodes, and time steps.

[0065] In some embodiments, the basic architecture of the LSTM neural network model can be an LSTM neural network model with 2 layers, 128 hidden nodes, and a time step of 5.

[0066] The electronic device uses the original battery samples of the original battery and the virtual battery samples of the generated virtual battery as the training set to construct a mapping model between HF and the battery capacity within the optimal voltage segment data.

[0067] Based on the LSTM neural network model obtained by the above method, in this embodiment, a method for estimating the battery capacity is provided, which is applied to an electronic device. See Figure 7 As shown, it includes the following steps S701 - S702.

[0068] S701. The electronic device obtains the charging data of the battery to be measured within the optimal voltage segment, where the sum of the correlations between the multiple health indicators corresponding to the charging data and the battery capacity is the highest among all voltage segments.

[0069] In this embodiment, the segment data of the battery to be measured within the optimal voltage segment includes the time, current, and voltage data generated by the battery to be measured within the optimal voltage segment during the constant current charging stage.

[0070] Exemplarily, if the original battery used for training the LSTM neural network model is the battery with model CS2-35 in CALCE, and after a series of calculations, the optimal voltage segment of this type of battery is 3.91V - 4.13V, then when detecting the battery capacity of the same type of batteries CS2-36 or CS2-37, the segment data selected is the time, current, and voltage data generated by the battery to be tested CS2-36 or CS2-37 within the voltage range of 3.91V - 4.13V.

[0071] Table 2 below shows the calculation results of some data of CALCE type batteries within the optimal voltage segment. As shown in Table 2, the optimal voltage segment 3.91V - 4.13V represents the optimal voltage range calculated for three batteries of the same type, CS2-35, CS2-36, and CS2-37. Within this optimal voltage range, the total score of the correlation between each type of HF and the battery capacity is the highest. The starting SOC and the ending SOC meet the preset constraint conditions. The average proportion of the segment charging time in the complete charging time represents the proportion of the charging duration within the optimal voltage segment compared to the charging duration of a full charge (i.e., fully charging and discharging) of the battery in the traditional technology when extracting data. For example, when fully charging and discharging the battery to be tested, the duration used is 10h, while the charging duration based on the optimal voltage segment may be 2.5h, greatly saving the time cost and improving the detection efficiency.

[0072] Table 2 Calculation Results of Some Data within the Optimal Voltage Segment

[0073]

[0074]

[0075] S702. The electronic device determines multiple health indicators of the charging data.

[0076] S703. The electronic device inputs the multiple health indicators into the trained neural network model to obtain the battery capacity of the battery to be tested, where the neural network model is trained through training samples, and the training samples are used to represent the corresponding relationship between the battery capacity and the multiple health indicators.

[0077] In an example of this embodiment, three batteries from the CALCE battery aging dataset are used: CS2-35 as the training set, and CS2-36 and CS2-37 as the test sets to implement and verify the method provided in the embodiments of the present application. HF is extracted from the optimal voltage segment of the CS2-35 battery, and 50 sets of virtual battery samples are generated as the training set to train the LSTM neural network model. And CS2-36 and CS2-37 are used as the test sets to verify the trained LSTM neural network model.

[0078] Figure 8 Schematic diagram of the estimated results of the CS2-36 battery capacity using the trained LSTM neural network model. See Figure 8 as shown Figure 8 The left figure in it is the estimated result without virtual battery samples, Figure 8 and the right figure in it is the estimated result with virtual battery samples. Figure 9 Schematic diagram of the estimated results of the CS2-37 battery capacity using the trained LSTM neural network model. See Figure 9 as shown Figure 9 The left figure in it is the estimated result without virtual battery samples, Figure 9 and the right figure in it is the estimated result with virtual battery samples. Among them, for battery CS2-36 and battery CS2-36, their Root Mean Square Error (RMSE), Mean Absolute Percentage Error (MAPE), and squared difference R 2 are calculated by the following formula.

[0079]

[0080]

[0081]

[0082] In the above formula, is the estimated capacity, y i is the actual battery capacity, is the average value of the actual battery capacity, and m is the number of data.

[0083] Table 3 Comparison of results with and without virtual battery samples

[0084]

[0085] Table 3 shows the calculation results of RMSE, MAPE, and R 2 in the case of having and not having virtual battery samples. As shown in Table 3, after adding virtual battery samples, the data changes of RMSE, MAPE, and R 2 show that the accuracy of the estimated results has been greatly improved, indicating the practicality of this method. Using three batteries in the CALCE battery dataset to verify the method proposed in this paper, the average RMSE of the test set is 1.64%. Therefore, the battery capacity estimation method provided by the embodiments of this application can accurately and quickly obtain the capacity of the battery.

[0086] In this embodiment, the electronic device for executing the LSTM neural network model construction method and the electronic device for executing the battery capacity estimation method may be the same electronic device or different electronic devices.

[0087] It should be understood that the sequence numbers of the steps in the above embodiments do not indicate the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present application.

[0088] Figure 10 is a schematic diagram of an electronic device provided by an embodiment of the present application. As Figure 10 shown, the electronic device 10 in this embodiment includes: a processor 100, a memory 101, and a computer program 102 stored in the memory 101 and executable on the processor 100, such as a battery capacity estimation program. When the processor 100 executes the computer program 102, the steps in the above embodiments of the battery capacity estimation method are implemented. Alternatively, when the processor 100 executes the computer program 102, the functions of each module / unit in the above device embodiments are implemented.

[0089] Exemplarily, the computer program 102 may be divided into one or more modules / units. The one or more modules / units are stored in the memory 101 and executed by the processor 100 to complete the present application. The one or more modules / units may be a series of computer program instruction segments capable of performing specific functions, and these instruction segments are used to describe the execution process of the computer program 102 in the electronic device 10.

[0090] The electronic device 10 may be a computing device such as a tablet computer, a tablet computer, a desktop computer, a notebook, a handheld computer, and a cloud server. The electronic device may include, but is not limited to, a processor 100 and a memory 101. Those skilled in the art can understand that Figure 10 merely examples of the electronic device 10 do not constitute a limitation to the electronic device 10, and it may include more or fewer components than shown in the figure, or combine certain components, or different components. For example, the electronic device may further include input / output devices, network access devices, a bus, etc.

[0091] The so-called processor 100 may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0092] The memory 101 may be an internal storage unit of the electronic device 10, such as the hard disk or memory of the electronic device 10. The memory 101 may also be an external storage device of the electronic device 10, such as a plug-in hard disk equipped on the electronic device 10, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. Further, the memory 101 may also include both the internal storage unit of the electronic device 10 and the external storage device. The memory 101 is used to store the computer program and other programs and data required by the electronic device. The memory 101 may also be used to temporarily store data that has been output or is to be output.

[0093] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the above division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules as needed, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiments can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction and do not limit the protection scope of this application. The specific working processes of the units and modules in the above system can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0094] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For the parts not detailed or recorded in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0095] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0096] In the embodiments provided in this application, it should be understood that the disclosed device / terminal device and method can be implemented in other ways. For example, the device / terminal device embodiments described above are only illustrative. For example, the division of the modules or units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces. The indirect couplings or communication connections of the device or unit can be in electrical, mechanical or other forms.

[0097] The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0098] In addition, the functional units in each embodiment of this application can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0099] When the integrated module / unit is implemented in the form of 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, to implement all or part of the processes in the above-described embodiment methods of the present application, it can also be completed by hardware related to computer program instructions. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-described various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0100] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for estimating battery capacity, characterized in that, The method includes: Obtaining charging data of the battery to be tested within a target voltage segment, where the sum of the correlations between multiple health indicators corresponding to the charging data and the battery capacity is the highest among all voltage segments. The target voltage segment is determined based on the actual aging data of the original battery and the types of the multiple health indicators. The original battery and the battery to be tested are of the same type; Determining the multiple health indicators of the charging data; Inputting the multiple health indicators into a trained neural network model to obtain the battery capacity of the battery to be tested. The neural network model is obtained by training with training samples, and the training samples are used to represent the corresponding relationship between the battery capacity and the multiple health indicators; Wherein, determining the target voltage segment based on the actual aging data and the types of the multiple health indicators includes: Determining multiple voltage segments based on the actual aging data; wherein, in each cycle of charge and discharge, each voltage segment has corresponding voltage segment data; According to the constraint conditions screen each of the voltage segment data; R is the voltage segment width, and the starting SOC is less than or equal to the ending SOC; Determine the F value of each of the filtered voltage segment data within each voltage segment, where r i is the score of the correlation between the i-th health indicator of the voltage segment data and the battery capacity; Determining the voltage segment corresponding to the maximum F value as the target voltage segment.

2. The method according to claim 1, wherein The multiple health indicators include M data domain health indicators and N model domain health indicators, M + N ≥ 2, M ≥ 0, N ≥ 0; The data domain health indicators include at least one of segment capacity, peak value of the incremental capacity analysis (ICA) curve, voltage corresponding to the peak value of the ICA curve, charging time, and voltage rise per unit time; The model domain health indicators include at least one of ohmic internal resistance, polarization internal resistance, polarization capacitance, and time constant.

3. The method according to claim 1 or 2, wherein The training samples include original battery samples and virtual battery samples, and the virtual battery samples are generated based on the original battery samples.

4. The method according to claim 3, wherein The virtual battery samples are generated by the following method: Obtaining the actual aging data of the original battery, where the actual aging data includes the change data of time, voltage, and current corresponding to each cycle of charge and discharge during the entire process of the battery capacity of the original battery decaying from 100% to 0%; Determining the target voltage segment based on the actual aging data and the types of the multiple health indicators; Generating original battery samples based on the charging data within the target voltage segment; Performing data expansion on the battery capacity and each health indicator in the original battery samples to generate the virtual battery samples.

5. The method according to claim 4, wherein The performing data expansion on the battery capacity and each health indicator in the original battery samples to generate the virtual battery samples includes: Determining the characteristic data center, left skewness, right skewness, and variance of the battery capacity and each health indicator in the original battery samples; Determining the upper expansion bound and lower expansion bound of the battery capacity and each health indicator based on the characteristic data center, the left skewness, the right skewness, and the variance; Performing data expansion on the battery capacity and each health indicator based on the upper expansion bound and lower expansion bound of the battery capacity and each health indicator to generate virtual battery samples that meet the preset conditions.

6. The method according to claim 1, wherein The neural network model is a long short-term memory (LSTM) neural network model.

7. The method according to claim 6, wherein The LSTM neural network model includes an LSTM neural network with 2 layers, 128 hidden nodes, and a time step of 5.

8. An electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method according to any one of claims 1 to 7.

Images