A lithium battery state of health estimation method, system, device and storage medium

Abstract

This invention discloses a method, system, device, and storage medium for estimating the health status of lithium batteries. The method includes: acquiring a pre-defined charge-discharge cycle dataset and analyzing it to obtain a pre-processed dataset; extracting IC, TIECVD, and DT features from the pre-processed dataset to obtain a health feature information set; randomly partitioning the health feature information set to obtain a training set and a test set; constructing a lithium battery health status prediction model based on an LSTM framework; inputting the training set into the lithium battery health status prediction model for training; and testing and validating the trained lithium battery health status prediction model using the test set to obtain a trained lithium battery health status prediction model; and estimating the health status of the lithium battery under test based on the trained lithium battery health status prediction model. This invention improves the accuracy of lithium battery health status estimation, thereby improving the safety and reliability of lithium battery operation, and can be applied in the field of lithium battery technology.

Description

Technical Field

[0001] This invention relates to the field of lithium battery technology, and in particular to a method, system, device, and storage medium for estimating the health status of lithium batteries. Background Technology

[0002] Lithium-ion batteries, as a clean energy source, are widely used in many fields, such as portable processing terminals, electric vehicles, and aerospace, due to their advantages of high energy density, low self-discharge rate, fast charging speed, and long lifespan. In practical applications, ensuring the safety and reliability of lithium-ion batteries is a crucial and complex issue. With continuous use and an increase in charge-discharge cycles, lithium-ion batteries undergo irreversible performance degradation, which can lead to malfunctions during operation and thus pose safety hazards. The State of Health (SOH) of a battery is an important parameter reflecting its current aging state. Battery aging is a gradual process influenced by various complex internal and external factors, such as internal chemical degradation, ambient temperature, and usage conditions. Battery aging significantly impacts the normal and safe operation of power systems. Predicting the SOH of a battery allows for assessment of its aging status, enabling timely battery replacement and preventing safety issues.

[0003] To avoid a series of problems caused by battery health issues, numerous studies on battery state of health (SOH) prediction have emerged in recent years. Temperature is closely related to lithium battery aging; for example, high temperatures accelerate internal chemical reactions and promote some adverse, irreversible chemical reactions. However, only a few studies have conducted in-depth analyses of the relationship between temperature and lithium battery aging to explore the connection between temperature and SOH. Most studies either do not use temperature or simply use surface temperature as a feature for SOH prediction, resulting in insufficient accuracy in estimating the state of health of lithium batteries. Summary of the Invention

[0004] The purpose of this invention is to at least partially solve one of the technical problems existing in the prior art.

[0005] Therefore, one objective of this invention is to provide a method for estimating the state of health of a lithium battery, which improves the accuracy of the estimation and thus enhances the safety and reliability of lithium battery operation.

[0006] Another objective of this invention is to provide a lithium battery health status estimation system.

[0007] To achieve the above-mentioned technical objectives, the technical solutions adopted in the embodiments of the present invention include:

[0008] On one hand, embodiments of the present invention provide a method for estimating the health status of a lithium battery, comprising the following steps:

[0009] Obtain a preset charge-discharge cycle dataset, and analyze and process the charge-discharge cycle dataset to obtain a preprocessed dataset;

[0010] IC, TIECVD, and DT features are extracted from the preprocessed dataset to obtain a health feature information set. The health feature information set is randomly divided, and the feature parameters of the IC, TIECVD, and DT feature extraction process are determined through comparative experiments to obtain a training set and a test set.

[0011] A lithium battery health status prediction model is constructed based on the LSTM framework. The training set is input into the lithium battery health status prediction model for training to obtain the trained lithium battery health status prediction model. The trained lithium battery health status prediction model is then tested and verified using the test set to obtain the trained lithium battery health status prediction model.

[0012] The health status of the lithium battery under test is estimated based on the trained lithium battery health status prediction model.

[0013] Furthermore, in one embodiment of the present invention, the step of analyzing and processing the charge-discharge cycle dataset to obtain a preprocessed dataset specifically includes:

[0014] The charge-discharge cycle dataset is cleaned to obtain the cleaned charge-discharge cycle dataset;

[0015] The cleaned charge-discharge cycle dataset is smoothed to obtain the preprocessed dataset.

[0016] Furthermore, in one embodiment of the present invention, the step of acquiring audio sample data and extracting IC, TIECVD, and DT features from the preprocessed dataset to obtain a health feature information set specifically includes:

[0017] The energy-time curve is obtained by calculating based on the current-time curve in the preprocessed dataset;

[0018] The IC curve is obtained by calculating based on the energy-time curve and the voltage-time curve in the preprocessed dataset;

[0019] The IC curve is smoothed using MASF filtering to obtain the IC curve characteristics;

[0020] Based on the charging time and terminal voltage in the preprocessed dataset, an equal voltage range charging time curve is obtained;

[0021] The charging time curves within the equal voltage range are sampled at equal intervals to obtain the TIECVD feature vector;

[0022] The DT curves in the preprocessed dataset are smoothed by SG filtering to obtain the smoothed DT curves;

[0023] Feature extraction is performed on the smoothed DT curve to obtain the DT feature vector;

[0024] The health feature information set is generated based on the IC curve features, the TIECVD feature vector, and the DT feature vector.

[0025] Furthermore, in one embodiment of the present invention, the step of constructing a lithium battery health state prediction model based on the LSTM framework specifically includes:

[0026] A forgetting gate is constructed based on sigmoid neural network layers and multiplication operation structure;

[0027] The input gate is constructed based on sigmoid neural network layers, tanh neural network layers, multiplication structures, and addition structures.

[0028] The output gate is constructed based on sigmoid neural network layers, tanh neural network layers, and addition operation structures.

[0029] Based on the LSTM framework, the lithium battery health status prediction model is constructed according to the forget gate, the input gate, and the output gate.

[0030] Furthermore, in one embodiment of the present invention, the calculation formula of the lithium battery health status prediction model is as follows:

[0031]

[0032] Where sigmoid(x) represents the operation of the sigmoid function on x, and tanh(x) represents the operation of the tanh function on x. t h represents the input at the current moment. t-1 h represents the output signal at the previous time step. t C represents the output signal at the current moment. t-1 This indicates the cell state at the previous time step. C represents spare information used to update the cell state. t f represents the current cell state. t i t o t W represents the outputs of the forget gate, input gate, and output gate at the current time step. f Wi W o b represents the weight parameters of the forget gate, input gate, and output gate, respectively. f b i and b o W represents the deviation parameters of the forget gate, input gate, and output gate, respectively. C b represents the weight parameter used to update the cell state. C This represents the deviation parameter used to update the cell state.

[0033] Furthermore, in one embodiment of the present invention, the step of inputting the training set into the lithium battery health status prediction model for training to obtain the trained lithium battery health status prediction model specifically includes:

[0034] The training set is input into the lithium battery health status prediction model to obtain the model output value;

[0035] The loss value is obtained by calculating the model output value and the true value of the training set using a preset loss function;

[0036] Based on the loss value, the model parameters of the lithium battery health status prediction model are optimized using the BP algorithm and gradient descent method. Then, the training set is input into the lithium battery health status prediction model until the preset number of training iterations is reached, at which point training stops, and the trained lithium battery health status prediction model is obtained.

[0037] Furthermore, in one embodiment of the present invention, the step of estimating the health status of the lithium battery under test based on the trained lithium battery health status prediction model specifically includes:

[0038] Obtain real-time charge and discharge information of the lithium battery under test;

[0039] The real-time charging and discharging information is input into the trained lithium battery health status prediction model to obtain the health status estimation result of the lithium battery under test.

[0040] On the other hand, embodiments of the present invention provide a lithium battery health status estimation system, comprising:

[0041] The dataset acquisition module is used to acquire a preset charge-discharge cycle dataset and analyze and process the charge-discharge cycle dataset to obtain a preprocessed dataset.

[0042] The feature extraction module is used to extract IC, TIECVD and DT features from the preprocessed dataset to obtain a health feature information set. The health feature information set is randomly divided, and the feature parameters of the IC, TIECVD and DT feature extraction process are determined by comparative experiments to obtain a training set and a test set.

[0043] The model training module is used to build a lithium battery health status prediction model based on the LSTM framework. The training set is input into the lithium battery health status prediction model for training to obtain the trained lithium battery health status prediction model. The trained lithium battery health status prediction model is tested and verified by the test set to obtain the trained lithium battery health status prediction model.

[0044] The health status estimation module is used to estimate the health status of the lithium battery under test based on the trained lithium battery health status prediction model.

[0045] On the other hand, embodiments of the present invention provide a lithium battery health status estimation device, which includes a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for implementing communication between the processor and the memory. When the program is executed by the processor, it implements the lithium battery health status estimation method as described above.

[0046] On the other hand, embodiments of the present invention also provide a storage medium, which is a computer-readable storage medium for computer-readable storage. The storage medium stores one or more programs, which can be executed by one or more processors to implement the lithium battery health state estimation method as described above.

[0047] The advantages and beneficial effects of the present invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention:

[0048] In this embodiment of the invention, a preset charge-discharge cycle dataset is obtained, and the dataset is analyzed and processed to obtain a preprocessed dataset. Then, IC, TIECVD, and DT features are extracted from the preprocessed dataset to obtain a health feature information set. The health feature information set is randomly divided, and the feature parameters of the IC, TIECVD, and DT feature extraction process are determined through comparative experiments to obtain a training set and a test set. A lithium battery health status prediction model is then constructed based on the LSTM framework. The training set is input into the lithium battery health status prediction model for training to obtain a trained lithium battery health status prediction model. The trained lithium battery health status prediction model is then tested and verified using the test set to obtain a well-trained lithium battery health status prediction model. Finally, the health status of the lithium battery under test can be estimated based on the well-trained lithium battery health status prediction model. In this embodiment of the invention, a charge-discharge cycle dataset is acquired and pre-analyzed. Feature extraction is performed on the IC curve, the charging time curve within the same voltage range, and the DT curve to obtain multi-dimensional health feature information. The extracted health feature information is then divided into time series to obtain health feature information with time-series information. After fusing the health feature information, it is input into a lithium battery health state prediction model built based on the LSTM framework for model training and testing, resulting in a trained lithium battery health state prediction model. This lithium battery health state prediction model, based on DT feature extraction, fully considers the relationship between temperature and lithium battery health state, improving the accuracy of lithium battery health state estimation, thereby improving the safety and reliability of lithium battery operation. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings used in the embodiments of the present invention are described below. It should be understood that the drawings described below are only for the convenience of clearly describing some embodiments of the technical solutions of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 A flowchart illustrating the steps of the lithium battery health status estimation method provided in this embodiment of the invention;

[0051] Figure 2 A schematic diagram of the IC curve provided in an embodiment of the present invention;

[0052] Figure 3 This is a schematic diagram of the charging time curve within the same voltage range provided in an embodiment of the present invention;

[0053] Figure 4 A schematic diagram of a DT curve provided in an embodiment of the present invention;

[0054] Figure 5Another schematic diagram of the DT curve provided in an embodiment of the present invention;

[0055] Figure 6 This is a schematic diagram of the structure of the lithium battery health status prediction model provided in an embodiment of the present invention;

[0056] Figure 7 A schematic diagram of the test results of the lithium battery health status prediction model provided in the embodiments of the present invention;

[0057] Figure 8 This is a schematic diagram of the structure of the lithium battery health status estimation system provided in an embodiment of the present invention;

[0058] Figure 9 This is a schematic diagram of the hardware structure of a lithium battery health status estimation device provided in an embodiment of the present invention. Detailed Implementation

[0059] The embodiments of the present invention are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application. It should be noted that although functional modules are divided in the system schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the system schematic diagram or the order in the flowchart. The step numbers in the following embodiments are only set for ease of explanation and do not limit the order between steps. The execution order of each step in the embodiments can be adaptively adjusted according to the understanding of those skilled in the art.

[0060] In the description of this invention, "multiple" means two or more. The use of "first" and "second" is for distinguishing technical features only and should not be construed as indicating or implying relative importance, the number of indicated technical features, or the order of the indicated technical features. Furthermore, unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0061] The lithium battery health state estimation method provided in this application can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, set-top box, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application that implements the lithium battery health state estimation method, but is not limited to the above forms.

[0062] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0063] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards of the relevant countries and regions. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirects to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data for the proper functioning of the embodiments of this application obtained.

[0064] like Figure 1 The diagram shown is a flowchart of one step of the lithium battery health state estimation method provided in an embodiment of the present invention. (Refer to...) Figure 1 This invention provides a method for estimating the health status of a lithium battery, specifically including the following steps:

[0065] S101. Obtain a preset charge-discharge cycle dataset and analyze and process the charge-discharge cycle dataset to obtain a preprocessed dataset.

[0066] Specifically, time-varying parameters are obtained, including the terminal voltage, charging current, temperature, and time at each moment in the lithium battery charging cycle. The obtained time-varying parameters are used as data for the corresponding charging cycle, and the data from multiple consecutive charging cycles are concatenated to obtain the charge-discharge cycle dataset.

[0067] As a further optional implementation, the step of analyzing and processing the charge-discharge cycle dataset to obtain a preprocessed dataset specifically includes:

[0068] S1011. Clean the charge-discharge cycle dataset to obtain the cleaned charge-discharge cycle dataset.

[0069] S1012. Smooth the cleaned charge-discharge cycle dataset to obtain a preprocessed dataset.

[0070] S102. Perform IC, TIECVD and DT feature extraction on the preprocessed dataset to obtain a health feature information set. Randomly divide the health feature information set and determine the feature parameters of the IC, TIECVD and DT feature extraction process through comparative experiments to obtain the training set and test set.

[0071] Specifically, for SOH prediction of lithium-ion batteries, a key step is to analyze the characteristics of battery measurement data to study how to effectively extract HFs. This involves observing the curves of lithium battery terminal voltage versus time under different charge-discharge cycles, the curves of charging current versus time under different charge-discharge cycles, and the curves of temperature versus time under different charge-discharge cycles. The IC, TIECVD, and DT features obtained after transforming the charging time, current, voltage, and temperature in a certain way are then selected as features for model training.

[0072] As an optional implementation, the step of extracting IC, TIECVD, and DT features from the preprocessed dataset to obtain a health feature information set specifically includes:

[0073] S1021. Calculate the charge-time curve based on the current-time curve in the preprocessed dataset;

[0074] S1022. Calculate the IC curve based on the energy-time curve and the voltage-time curve in the preprocessed dataset;

[0075] S1023. Perform MASF filtering smoothing on the IC curve to obtain the IC curve characteristics;

[0076] S1024. Based on the charging time and terminal voltage in the preprocessed dataset, obtain the charging time curve for the equal voltage range.

[0077] S1025. Perform equidistant sampling on the charging time curve of the equal voltage range to obtain the TIECVD feature vector;

[0078] S1026. Perform SG filtering on the DT curves in the preprocessed dataset to obtain smoothed DT curves.

[0079] S1027. Extract features from the smoothed DT curve to obtain the DT feature vector;

[0080] S1028. Generate a health feature information set based on the IC curve features, TIECVD feature vector, and DT feature vector.

[0081] Specifically, the process of feature extraction for IC, TIECVD, and DT will be explained below.

[0082] 1) IC feature extraction:

[0083] The capacitance-voltage curve, differentiated with respect to voltage, yields the capacitance-discharge (IC) curve, which illustrates the relationship between charge / discharge capacity and voltage. For example... Figure 2 The figure shown is a schematic diagram of the IC curve provided in an embodiment of the present invention. Figure 2 It can be seen that the IC curves under different charge-discharge cycles exhibit significant differences in the region near the peak. Features extracted from the IC curves can reveal the relationship between voltage and battery capacity. The calculation expression for the IC curve is shown below:

[0084]

[0085] In the formula, Q k η is the total amount of electricity charged up to time k, I is the charging efficiency, t is the charging current, and V is the terminal voltage. From equation (2), it can be seen that when the sampling interval is small, the denominator V... k+1 -V k A small value will cause significant fluctuations in the calculation results. In practical applications, issues such as sensor noise and equipment instability can cause fluctuations in the measured voltage, which in turn will affect V. k+1 -V kOutliers, such as zero or negative values, can cause anomalies in the calculation results and are detrimental to subsequent data processing. This problem can be addressed using smoothing filtering algorithms, such as Kalman filtering and Gaussian filtering. To simplify the calculation, we choose to use the Moving Average Smoothing Filter (MASF), as shown in the following formula:

[0086]

[0087] Where x is the original input, y is the processed output, and M is the size of the smoothing window. Similar to the mean filtering algorithm, MASF essentially replaces the value of the element at the current position with the mean of the M neighboring elements.

[0088] The specific steps for smoothing the IC curve and extracting IC features using MASF filtering are as follows:

[0089] ① Calculate the charging capacity Q at all times based on the current curve;

[0090] ② Select window size M1, and use MASF to smooth the voltage V and the charging capacity Q respectively;

[0091] ③ Calculate dQ / dV, select ΔV = 0.1mV as the unit of voltage V, then perform interpolation on voltage V, i.e., expand the data of voltage V, and then perform the same operation on dQ / dV. This process ensures that for any voltage point (the smallest unit should not be less than 0.1mV, such as 3.8625V), dQ / dV can be obtained with the corresponding value;

[0092] ④ Select window size M2, and use MASF to smooth the dQ / dV obtained in step ③. The dQ / dV-V curve at this time is the smoothed IC curve.

[0093] ⑤ Select a voltage range V1~V2 and a sampling point number N. Perform equidistant sampling on the IC curve obtained in step ④ to obtain the IC feature vector: [IC1, IC2, ..., IC2]. N ].

[0094] 2) TIECVD Feature Extraction:

[0095] During the constant current charging phase, the relationship between the charging time required for the battery's terminal voltage to reach the cutoff voltage of 4.2V under different charge-discharge cycles and the terminal voltage (i.e., the charging time curve within the equal voltage range) is as follows: Figure 3As shown, during the constant current charging stage, the battery's terminal voltage rises faster and the time to reach the cutoff voltage gradually shortens with the increase in the number of charge-discharge cycles. This indicates that the battery's full charge time is gradually shortening, meaning that the battery capacity is gradually decreasing. Therefore, the TIECVD feature extracted based on the charging time curve within the equal voltage range can effectively characterize the battery's capacity degradation characteristics.

[0096] The specific steps for extracting TIECVD features from the charging time curve within the same voltage range are as follows:

[0097] ① Based on the battery's measurement data, including charging time t and terminal voltage V, find the time t when the terminal voltage first reaches the cutoff voltage of 4.2V. e Then, t before the cutoff voltage i Terminal voltage V at time i The corresponding charging time for the equal voltage range is t. e -t i ;

[0098] ②Calculate the equal voltage range charging time corresponding to the terminal voltage at any moment before the cutoff voltage using the calculation method in step ①, and you can get the equal voltage range charging time curve.

[0099] ③ Select a voltage range V1~V2 and a sampling point number N. Perform equidistant sampling on the charging time curve of the equal voltage range obtained in step ② to obtain the TIECVD feature vector: [TIECVD1, TIECVD2, ..., TIECVD... N ].

[0100] 3) DT Feature Extraction:

[0101] The formulas for the DT physics model are shown below:

[0102]

[0103] Where T is temperature, I is current, and R... i Let F be the battery internal resistance, n be the number of exchanged electrons, ΔS be the entropy change, and m be the number of electrons exchanged. c A and C represent the mass area and surface area of ​​the tested battery, respectively. p and h are the heat capacity and equivalent heat transfer coefficient, respectively, T amb For ambient temperature. 2 R i This indicates Joule heating caused by the battery's internal resistance. hA(TT) represents the reversible heat of reaction caused by the entropy change at the electrode during lithium-ion insertion or extraction. amb The symbol () represents the heat exchange between the battery's interior and the environment. During constant current charging, only R in equation (3.4) is considered. iWhile ΔS changes, other parameters remain constant. In this case, the change in DT is only related to R. i It is related to ΔS. Since the changes in the battery's internal resistance and entropy are closely related to battery aging, the DT curve is closely related to battery aging. Extracting features based on the DT curve can effectively characterize battery aging characteristics.

[0104] Because temperature sensors have limited resolution, direct calculation of DT is easily affected by measurement noise. To avoid this problem, the DT curve is approximately calculated using the following formula, where L is the sampling interval.

[0105]

[0106] The temperature curve (DT) of a lithium battery under different charge-discharge cycles contains a lot of noise when it is not processed. It is difficult to extract features from the DT curve with a lot of noise. Therefore, filtering algorithms such as Kalman filtering and Gaussian filtering can be used to smooth the temperature curve and the DT curve at the same time.

[0107] To reduce computational load, this study uses the SG filtering algorithm. When the temperature curve contains a lot of noise, it needs to be smoothed first, then the smoothed temperature curve is used to calculate the DT curve, and then the DT curve is smoothed again to obtain the final DT curve. When the temperature curve has less noise, only the DT curve needs to be smoothed. The specific steps are as follows:

[0108] ① The original temperature curve is smoothed using SG filtering, with the sampling interval L selected. A The DT curve is calculated using the smoothed temperature curve according to equation (5);

[0109] ② Apply SG filtering to the DT curve obtained in step ① to obtain the smoothed DT curve;

[0110] ③ For example Figure 4 The diagram shown illustrates a feature extraction method for the DT curve according to an embodiment of the present invention. Specifically, it extracts the amplitude H1 of peak 1, the amplitude H2 of trough 2, the amplitude H3 of peak 2, the amplitude difference H4 between peak 1 and trough 2, the amplitude difference H5 between peak 2 and trough 2, and the voltage value V of peak 1. A The voltage value V at trough 2 B Voltage value V of peak 2 C The voltage difference D1 between peak 1 and trough 2, and the voltage difference D2 between peak 2 and trough 2, such as Figure 5The diagram illustrates another method for feature extraction of the DT curve according to an embodiment of the present invention. Specifically, it extracts the area A1 enclosed by the curve from trough 1 to peak 1, the horizontal line containing the amplitude of trough 1, and the vertical line containing the voltage value of peak 1; the area A2 enclosed by the curve from peak 1 to trough 2, the horizontal line containing the amplitude of trough 2, and the vertical line containing the voltage value of peak 1; and the area A3 enclosed by the curve from trough 2 to peak 2, the horizontal line containing the amplitude of trough 2, and the vertical line containing the voltage value of peak 2. The DT curve obtained in step ② is then used... Figure 3 and Figure 4 The feature extraction method shown selects multiple health features highly correlated with SOH from the smoothed DT curve to obtain the DT feature vector, for example [V A V C H4, D1, D2, A1, A2).

[0111] In some optional embodiments, after extracting the IC curve features, TIECVD feature vector, and DT feature vector, to avoid weight skew due to differences in feature units and orders of magnitude, the IC, TIECVD, and DT features need to be dimensionless processed, such as normalization, standardization, and unit scaling, to ensure that the orders of magnitude of all features remain consistent. The formulas for normalization and standardization are shown below:

[0112]

[0113] in:

[0114]

[0115] Since the dimensions of the IC, TIECVD, and DT feature vectors are inconsistent, and it is not necessary for them to be consistent, weight coefficients can be assigned to each feature vector before vector concatenation. These weight coefficients can be obtained manually or calculated using algorithms such as the β distribution. To simplify the algorithm, this embodiment of the invention uses a manual assignment method to define the weight coefficients.

[0116] In some optional embodiments, this embodiment uses the hold-out method, randomly dividing the dataset into a training set and a test set in a 3:1 ratio. The training set is then used to train the model, and the model's effectiveness and generalization ability can be determined by evaluating its performance on the test set, allowing for necessary adjustments and improvements.

[0117] In some optional embodiments, the present invention determines the parameters of the IC, TIECVD, and DT feature extraction processes through comparative experiments on a dataset. The specific process is as follows:

[0118] 1) Selection of the number of IC feature points:

[0119] In this embodiment of the invention, the voltage range of the region where the IC curve of the battery differs significantly under different charge-discharge cycles is 3.5V to 4.1V. This voltage range is selected as the interval for extracting IC features. Using the IC feature extraction method described in the preceding steps, the voltage range is 3.5V to 4.2V, and the number of sampling points N is 5, 10, 15, and 20, respectively.

[0120] In this embodiment of the invention, the number of sampling points N for the IC features used is determined to be 10.

[0121] 2) Selection of the number of TIECVD feature points:

[0122] In this embodiment of the invention, the voltage range of the region where the charging time curve of the battery varies significantly under different charge-discharge cycles is 3.2V to 4.1V. This voltage range is selected as the interval for extracting TIECVD features. Using the TIECVD feature extraction method described in the preceding steps, the voltage range is 3.2V to 4.1V, and the number of sampling points N is 5, 10, 15, and 20, respectively.

[0123] In this embodiment of the invention, the number of sampling points N for the TIECVD feature used is determined to be 10.

[0124] 3) DT Feature Selection:

[0125] The DT feature extraction method involves V features. B V C V D To determine which features to use, H1, H2, H3, H4, H5, D1, D2, A1, A2, and A3, Pearson correlation analysis can be used. The Pearson correlation coefficient is independent of the dimensions or orders of magnitude of the data and is typically used to measure the degree of linear correlation between two sets of data. The Pearson correlation coefficient ranges from -1 to 1. The larger the absolute value of the correlation coefficient, the stronger the linear correlation between the two sets of data; for example, 1 indicates a positive linear correlation, -1 indicates a negative linear correlation, and 0 indicates no linear correlation. The calculation formula is shown below:

[0126]

[0127] in:

[0128]

[0129] Where r xy Let x represent the Pearson correlation coefficient between vectors x and y, and n represent the lengths of vectors x and y. i and y i Let these represent the values ​​of the i-th elements of vectors x and y, respectively. and Let x and y represent the mean values ​​of vectors x and y, respectively. In the analysis of DT features, vector x corresponds to a vector composed of a certain feature under different charge-discharge cycles, and vector y corresponds to a vector composed of SOH under different charge-discharge cycles.

[0130] The feature finally selected in this embodiment of the invention is V. B V D H1, H4, A2 and A3.

[0131] 4) Selection of the length for dividing the time series

[0132] The size of the feature matrix input to the LSTM network is (samples, sequence_len, features_dim), where samples represents the number of samples, which in this embodiment represents the number of different charge-discharge cycles, sequence_len represents the time series length of each sample, and features_dim represents the number of features.

[0133] For lithium batteries, each charge-discharge cycle can only extract one feature vector of length features_dim (the feature vector X of the i-th charge-discharge cycle). i =[X i，1 X i,2 , ......, X i，features_dim If the feature matrix of each battery is X = [X1, X2, ..., X...], then the feature matrix of each battery is X = [X1, X2, ..., X...]. C ], where C is the number of charge-discharge cycles. Therefore, it is necessary to divide the time series, that is, to divide X into [X... 1：sequence_len X 2：sequence_len+1 , ......, X C-sequence_len+1：C ].

[0134] The time series lengths were set to 5, 10, 15, and 20, respectively, and the optimal feature parameters obtained earlier were used to extract IC, TIECVD, and DT features as inputs to the model for experiments.

[0135] In this embodiment of the invention, the time series length used is determined to be 10.

[0136] S103. Construct a lithium battery health status prediction model based on the LSTM framework, input the training set into the lithium battery health status prediction model for training, obtain the trained lithium battery health status prediction model, and test and verify the trained lithium battery health status prediction model through the test set to obtain the trained lithium battery health status prediction model.

[0137] Specifically, since lithium battery measurement data is related to the number of charge-discharge cycles and has information on a time scale, and the extracted HFs also have information on a time scale, LSTM networks can be used, and the problems of gradient vanishing and gradient exploding when dealing with long-term memory can be avoided.

[0138] As an optional implementation, the step of constructing a lithium battery health status prediction model based on the LSTM framework specifically includes:

[0139] S1031. Construct a forget gate based on sigmoid neural network layers and multiplication operation structure;

[0140] S1032. Construct an input gate based on a sigmoid neural network layer, a tanh neural network layer, a multiplication structure, and an addition structure;

[0141] S1033. Construct an output gate based on a sigmoid neural network layer, a tanh neural network layer, and an addition operation structure;

[0142] S1034. Based on the LSTM framework, a lithium battery health status prediction model is constructed according to the forget gate, input gate, and output gate.

[0143] Specifically, the core of LSTM is the cell state, denoted as C. t This is used to store the current LSTM state information and pass it to the LSTM at the next time step. The current LSTM receives the cell state C from the previous time step. t-1 The input x received by the current LSTM t The combined effect generates the current LSTM cell state C. t and C t The LSTM is passed to the next time step.

[0144] like Figure 6 The diagram shown illustrates the structure of a lithium battery health state prediction model provided in an embodiment of the present invention. The LSTM-based lithium battery health state prediction model includes a forget gate, a memory gate, and an output gate; wherein the forget gate is constructed from a sigmoid neural network layer and a bitwise multiplication structure, determining the cell state C. t-1 Which information is forgotten in the input? The memory gate is constructed from the input gate, tanh neural network layer, bitwise multiplication structure, and bitwise addition structure. The role of the memory gate is to determine the new input information x. t and h t-1 Which information will be retained? The output gate is constructed from a stgmoid neural network layer, a tanh neural network layer, and a bitwise addition structure. The output gate will take the unit state C passed from time t-1 after passing through the forget gate and the memory gate. t-1 The output signal h at time t-1 t-1and the input signal x at time t t These are combined and used as the output signal at the current moment. Where x... t and h t-1 The output after passing through a sigmoid neural network layer is a value between 0 and 1. t C t After passing through a tanh function, and at o t Multiply to obtain the output signal h t At the same time h t This signal is also passed to the next stage as the input signal for the next time step. Multiple LSTM networks can be stacked, and their output is then passed through a fully connected (FC) layer to map the learned feature information of the LSTM network into the sample label space. The final output is the predicted value of the remaining capacity of the lithium battery. In this embodiment, the number of LSTM network layers (num_layers) is 5, and the feature dimension (hidden_size) of the hidden layer is 50.

[0145] As an optional implementation, the calculation formula for the lithium battery health status prediction model is as follows:

[0146]

[0147] Where sigmoid(x) represents the operation of the sigmoid function on x, and tanh(x) represents the operation of the tanh function on x. t h represents the input at the current moment. t-1 h represents the output signal at the previous time step. t C represents the output signal at the current moment. t-1 This indicates the cell state at the previous time step. C represents spare information used to update the cell state. t f represents the current cell state. t i t o t W represents the outputs of the forget gate, input gate, and output gate at the current time step. f W i W o b represents the weight parameters of the forget gate, input gate, and output gate, respectively. f b i and b o W represents the deviation parameters of the forget gate, input gate, and output gate, respectively. C b represents the weight parameter used to update the cell state. C This represents the deviation parameter used to update the cell state.

[0148] As an optional implementation, the step of inputting the training set into the lithium battery health status prediction model for training, to obtain the trained lithium battery health status prediction model, specifically includes:

[0149] S1035. Input the training set into the lithium battery health status prediction model to obtain the model output value;

[0150] S1036. Calculate the loss value by using the preset loss function to calculate the model output value and the true value of the training set;

[0151] S1037. Based on the loss value, the model parameters of the lithium battery health status prediction model are optimized using the BP algorithm and gradient descent method. Then, the step of inputting the training set into the lithium battery health status prediction model is returned until the preset number of training iterations is reached, at which point training is stopped, and the trained lithium battery health status prediction model is obtained.

[0152] Specifically, a dimensionless scaling method with orders of magnitude scaling is used. The weight coefficients for feature fusion are IC:TIECVD:DT = 1:1:2. The voltage range for extracting IC features is 3.8V to 4.1V, with 10 sampling points. The voltage range for extracting TIECVD features is 3.4V to 4.2V, with 10 sampling points. The training set is input into the lithium battery health status prediction model. The mean squared error loss function is used to calculate the error value between the model output value and the true value of the training set. The calculation formula is as follows:

[0153]

[0154] Where MSE represents the mean squared error. This represents the model output value, y. i This represents the true values ​​in the training set, where n is the number of samples.

[0155] The BP algorithm, consisting of two processes—forward propagation of information and backward propagation of error—optimizes the model parameters to minimize the error, ultimately yielding a trained lithium battery health status prediction model.

[0156] In some optional embodiments, after 5 training cycles, the test dataset is input into the trained lithium battery health status prediction model for testing, and the test results are obtained.

[0157] This invention uses K-fold cross-validation, which involves averaging the error of K validation trials, to evaluate the model's predictive performance. The root mean square error, mean square error, and mean absolute percentage error (MSE) between the predicted and actual values ​​are calculated as performance metrics for the current model. Smaller errors indicate better model performance. The formulas for calculating these performance metrics are as follows:

[0158] Root mean square error:

[0159] Mean absolute error:

[0160] Mean absolute percentage error:

[0161] in, This is the predicted SOH value, y i This is the actual SOH value, and n is the number of charge-discharge cycles.

[0162] This invention employs a dimensionless scaling method with orders of magnitude scaling. The weight coefficients for feature fusion are IC:TIECVD:DT = 1:1:2. The voltage range for extracting IC features is 3.8V–4.1V, with 10 sampling points. The voltage range for extracting TIECVD features is 3.4V–4.2V, with 10 sampling points. The final trained model's test results on the test set are as follows: Figure 7 As shown, the errors of RMSE, MAE, and MAPE using only IC and TIECVD features are 0.748%, 0.574%, and 0.692%, respectively, while the errors using IC, TIECVD, and DT features simultaneously are 0.623%, 0.499%, and 0.599%, respectively.

[0163] S104. Estimate the health status of the lithium battery under test based on the trained lithium battery health status prediction model.

[0164] Specifically, the lithium battery health status prediction model trained based on the aforementioned steps can estimate the health status of the lithium battery under test. By inputting the real-time charging and discharging information of the lithium battery under test into the lithium battery health status prediction model, the health status estimation result output by the model can be obtained.

[0165] As a further optional implementation, the step of estimating the health status of the lithium battery under test based on the trained lithium battery health status prediction model specifically includes:

[0166] S1041. Obtain real-time charging and discharging information of the lithium battery under test;

[0167] S1042. Input the real-time charging and discharging information into the trained lithium battery health state prediction model to obtain the health state estimation result of the lithium battery under test.

[0168] The method steps of the embodiments of the present invention have been described above. It can be understood that the embodiments of the present invention acquire charge-discharge cycle datasets and perform pre-analysis processing, extracting features from IC curves, equal-voltage range charging time curves, and DT curves to obtain multi-dimensional health feature information. The extracted health feature information is then divided into time series to obtain health feature information with temporal information. This health feature information is then fused and input into a lithium battery health state prediction model built based on the LSTM framework for model training and testing, resulting in a trained lithium battery health state prediction model. This lithium battery health state prediction model, based on DT feature extraction, fully considers the relationship between temperature and lithium battery health state, improving the accuracy of lithium battery health state estimation, thereby improving the safety and reliability of lithium battery operation.

[0169] like Figure 8 The diagram shown is a structural schematic of the lithium battery health status estimation system provided in an embodiment of the present invention. (Refer to...) Figure 8 This invention provides a lithium battery health status estimation system, comprising:

[0170] The dataset acquisition module is used to acquire a preset charge-discharge cycle dataset and analyze and process the charge-discharge cycle dataset to obtain a preprocessed dataset.

[0171] The feature extraction module is used to extract IC, TIECVD and DT features from the preprocessed dataset to obtain a health feature information set. The health feature information set is randomly divided, and the feature parameters of the IC, TIECVD and DT feature extraction process are determined through comparative experiments to obtain the training set and the test set.

[0172] The model training module is used to build a lithium battery health status prediction model based on the LSTM framework. The training set is input into the lithium battery health status prediction model for training to obtain the trained lithium battery health status prediction model. The trained lithium battery health status prediction model is then tested and verified using the test set to obtain the trained lithium battery health status prediction model.

[0173] The health status estimation module is used to estimate the health status of the lithium battery under test based on the trained lithium battery health status prediction model.

[0174] The content of the above method embodiments is applicable to this system embodiment. The specific functions implemented in this system embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0175] This invention also provides a lithium battery health status estimation device, which includes: a memory, a processor, a program stored in the memory and executable on the processor, and a data bus for communication between the processor and the memory. When the program is executed by the processor, it implements the aforementioned lithium battery health status estimation method. This lithium battery health status estimation device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0176] like Figure 9 The diagram shown is a hardware structure schematic of the lithium battery health status estimation device provided in an embodiment of the present invention. (Refer to...) Figure 9 This invention provides a lithium battery health status estimation device, comprising:

[0177] The processor 901 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present invention.

[0178] The memory 902 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 902 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 902 and is called and executed by the processor 901 to execute the lithium battery health status estimation method of the embodiments of this invention.

[0179] The input / output interface 903 is used to implement information input and output;

[0180] The communication interface 904 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0181] Bus 905 transmits information between various components of the device (e.g., processor 901, memory 902, input / output interface 903, and communication interface 904);

[0182] The processor 901, memory 902, input / output interface 903, and communication interface 904 are connected to each other within the device via bus 905.

[0183] This invention also provides a storage medium, which is a computer-readable storage medium for computer-readable storage. The storage medium stores one or more programs, which can be executed by one or more processors to implement the above-described lithium battery health state estimation method.

[0184] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0185] This invention also discloses a computer program product or computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device can read the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, causing the computer device to perform... Figure 1 The method shown.

[0186] In some alternative embodiments, the functions / operations mentioned in the block diagrams may not occur in the order shown in the operation diagrams. For example, depending on the functions / operations involved, two consecutively shown blocks may actually be executed substantially simultaneously, or the aforementioned blocks may sometimes be executed in reverse order. Furthermore, the embodiments presented and described in the flowcharts of this invention are provided by way of example to provide a more comprehensive understanding of the technology. The disclosed methods are not limited to the operations and logic flows presented herein. Alternative embodiments are contemplated in which the order of various operations is changed and sub-operations described as part of a larger operation are executed independently.

[0187] Furthermore, although the invention has been described in the context of functional modules, it should be understood that, unless otherwise stated, one or more of the aforementioned functions and / or features may be integrated into a single physical device and / or software module, or one or more functions and / or features may be implemented in a separate physical device or software module. It is also understood that a detailed discussion of the actual implementation of each module is unnecessary for understanding the invention. Rather, given the properties, functions, and internal relationships of the various functional modules in the apparatus disclosed herein, the actual implementation of the module will be understood within the scope of conventional skill of an engineer. Therefore, those skilled in the art can implement the invention as set forth in the claims using ordinary techniques without excessive experimentation. It is also understood that the specific concepts disclosed are merely illustrative and not intended to limit the scope of the invention, which is determined by the full scope of the appended claims and their equivalents.

[0188] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a 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 described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0189] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0190] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the aforementioned program can be printed, because the aforementioned program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or, if necessary, processing in other suitable ways, and then stored in computer memory.

[0191] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0192] In the foregoing description of this specification, references to terms such as "one embodiment," "another embodiment," or "some embodiments" indicate that a specific feature, structure, material, or characteristic described in connection with an embodiment or example is included in at least one embodiment or example of the present invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0193] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

[0194] The above is a detailed description of the preferred embodiments of the present invention. However, the present invention is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or substitutions without departing from the spirit of the present invention. All such equivalent modifications or substitutions are included within the scope defined by the claims of this application.

Claims

1. A lithium battery state of health estimation method, characterized by, The method comprises the following steps: obtaining a preset charging and discharging cycle data set and performing analysis and processing on the charging and discharging cycle data set to obtain a pretreatment data set; performing IC, TIECVD and DT feature extraction on the pretreatment data set to obtain a health feature information set, performing random division on the health feature information set, and determining feature parameters of the IC, TIECVD and DT feature extraction process through a comparative experiment to obtain a training set and a test set; constructing a lithium battery health state prediction model based on an LSTM framework, inputting the training set into the lithium battery health state prediction model for training to obtain a trained lithium battery health state prediction model, and testing and verifying the trained lithium battery health state prediction model through the test set to obtain a trained lithium battery health state prediction model; estimating the health state of a lithium battery to be tested according to the trained lithium battery health state prediction model; the method of determining the feature parameters of the IC, TIECVD and DT feature extraction process through a comparative experiment specifically comprises: determining a first voltage range of a region with a difference greater than a preset first amplitude in IC curves under different charging and discharging cycles on the pretreatment data set through a comparative experiment, determining an IC feature extraction interval according to the first voltage range, and determining the number of feature points of the IC feature extraction process according to the IC feature extraction interval; determining a second voltage range of a region with a difference greater than a preset second amplitude in the constant-voltage-range charging time curve under different charging and discharging cycles on the pretreatment data set through a comparative experiment, determining a TIECVD feature extraction interval according to the second voltage range, and determining the number of feature points of the TIECVD feature extraction process according to the TIECVD feature extraction interval; determining a plurality of DT curves and SOH curves under different charge-discharge cycles through comparative experiments on the pretreatment data set, and determining a first amplitude of a first peak, a second amplitude of a second valley, a third amplitude of a second peak, a first amplitude difference between the first peak and the second valley, a second amplitude difference between the second peak and the second valley, a first voltage value of the first peak, a second voltage value of the second valley, a third voltage value of the second peak, a first voltage difference between the first peak and the second valley, a second voltage difference between the second peak and the second valley, a first area surrounded by a curve from the first valley to the first peak, a horizontal line at the amplitude of the first valley and a vertical line at the first voltage value of the first peak, a second area surrounded by a curve from the first peak to the second valley, a horizontal line at the second amplitude of the second valley and a vertical line at the first voltage value of the first peak, and a third area surrounded by a curve from the second valley to the second peak, a horizontal line at the second amplitude of the second valley and a vertical line at the third voltage value of the second peak according to the DT curves; determining a DT feature vector corresponding to each of the candidate DT features, and determining an SOH feature vector according to the SOH curve; determining a correlation coefficient of each of the DT feature vectors and the SOH feature vector through Pearson correlation analysis, and screening a target DT feature for constructing a training set and a data set from the candidate DT features according to the correlation coefficient.

2. The method of claim 1, wherein, The step of analyzing and processing the charge-discharge cycle data set to obtain a pretreatment data set specifically includes: performing data cleaning on the charge-discharge cycle data set to obtain a cleaned charge-discharge cycle data set; performing smoothing processing on the cleaned charge-discharge cycle data set to obtain the pretreatment data set.

3. The method of claim 1, wherein, The step of extracting IC, TIECVD and DT features from the pretreatment data set to obtain a health feature information set specifically includes: calculating an electric quantity-time curve according to a current-time curve in the pretreatment data set; calculating an IC curve based on the electric quantity-time curve and a voltage-time curve in the pretreatment data set; performing MASF filtering and smoothing processing on the IC curve to obtain an IC curve feature; obtaining an equal-voltage range charging time curve according to a charging time and an end voltage in the pretreatment data set; performing equidistant sampling on the equal-voltage range charging time curve to obtain a TIECVD feature vector; performing S-G filtering and smoothing processing on a DT curve in the pretreatment data set to obtain a smoothed DT curve; performing feature extraction on the smoothed DT curve to obtain a DT feature vector; Generate the health feature information set according to the IC curve feature, the TIECVD feature vector and the DT feature vector.

4. The method of claim 1, wherein, The step of constructing a lithium battery health state prediction model based on the LSTM framework specifically includes: Constructing a forget gate based on a sigmoid neural network layer and a multiplication operation structure; Constructing an input gate based on a sigmoid neural network layer, a tanh neural network layer, a multiplication operation structure and an addition operation structure; Constructing an output gate based on a sigmoid neural network layer, a tanh neural network layer and an addition operation structure; Constructing the lithium battery health state prediction model according to the forget gate, the input gate and the output gate based on the LSTM framework.

5. The method of claim 4, wherein, The operation formula of the lithium battery health state prediction model is as follows: wherein, denotes an operation of a function, denotes an operation of a function, denotes an input at a current time, denotes an output signal at a previous time, denotes an output signal at a current time, denotes a cell state at a previous time, denotes backup information for updating the cell state, denotes a cell state at a current time, , , denote outputs of a forget gate, an input gate, and an output gate at a current time, respectively, , , denote weight parameters of the forget gate, the input gate, and the output gate, respectively, , and denote bias parameters of the forget gate, the input gate, and the output gate, respectively, denotes a weight parameter for updating the cell state, denotes a bias parameter for updating the cell state.​​ 6. The method of claim 1, wherein, The step of inputting the training set into the lithium battery health state prediction model for training to obtain the trained lithium battery health state prediction model specifically includes: Inputting the training set into the lithium battery health state prediction model to obtain a model output value; Calculating the model output value and the true value of the training set by using a preset loss function to obtain a loss value; Optimizing the model parameters of the lithium battery health state prediction model by a BP algorithm and a gradient descent method based on the loss value, and returning the step of inputting the training set into the lithium battery health state prediction model until a preset training number is reached, stopping training and obtaining the trained lithium battery health state prediction model.

7. The lithium battery state of health estimation method of any one of claims 1 to 6, wherein, The step of estimating the health state of a to-be-tested lithium battery according to the trained lithium battery health state prediction model specifically includes: Obtaining real-time charge and discharge information of the to-be-tested lithium battery; Inputting the real-time charge and discharge information into the trained lithium battery health state prediction model to obtain a health state estimation result of the to-be-tested lithium battery.

8. A lithium battery state of health estimation system, characterized by, A lithium battery health state estimation method for implementing any one of claims 1 to 7 includes: A data set acquisition module for acquiring a preset charge and discharge cycle data set and analyzing and processing the charge and discharge cycle data set to obtain a pretreated data set; A feature extraction module for performing IC, TIECVD and DT feature extraction on the pretreated data set to obtain a health feature information set, randomly dividing the health feature information set, and determining feature parameters of the IC, TIECVD and DT feature extraction process through a comparative experiment to obtain a training set and a test set; A model training module for constructing a lithium battery health state prediction model based on an LSTM framework, inputting the training set into the lithium battery health state prediction model for training to obtain a trained lithium battery health state prediction model, and testing and verifying the trained lithium battery health state prediction model through the test set to obtain a trained lithium battery health state prediction model; A health state estimation module for estimating the health state of a to-be-tested lithium battery according to the trained lithium battery health state prediction model.

9. A lithium battery state of health estimation device, characterized by, The lithium battery state of health estimation device comprises a memory, a processor, a program stored on the memory and executable on the processor, and a data bus for realizing connection communication between the processor and the memory, and the program, when executed by the processor, realizes the steps of the lithium battery state of health estimation method according to any one of claims 1 to 7.

10. A storage medium, the storage medium being a computer-readable storage medium for computer-readable storage, characterized in that, The storage medium stores one or more programs, and the one or more programs are executable by one or more processors to realize the steps of the lithium battery state of health estimation method according to any one of claims 1 to 7.

Images