Lithium ion battery health state evaluation method and system

Abstract

The invention provides a lithium ion battery health state evaluation method and system, and the method comprises the steps: obtaining tangent slopes of different voltage points in a lithium ion battery IC curve, voltage values corresponding to extreme points, constant-current charging and discharging current features and constant-voltage charging voltage features, and taking the features as key features of a to-be-detected lithium ion battery; carrying out correlation calculation by adopting a Pearson's correlation coefficient, a Spearman level correlation coefficient and a maximum information coefficient to obtain a final health assessment factor for assessment; the extracted key features can effectively reflect the health state degradation performance of the battery, and the feature noise can be effectively reduced by adopting three different correlation methods for screening, so that the accuracy of later evaluation is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of lithium-ion battery health status assessment, and in particular relates to a lithium-ion battery health status assessment method and system. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] Lithium-ion batteries are widely used in electric vehicles, mobile devices, and grid energy storage systems. As the working time increases, the battery capacity and performance will inevitably decline, which will significantly affect the reliability and safety of the equipment. Therefore, accurately assessing the state of health (SOH) of lithium-ion batteries is crucial to optimize the battery management system, extend battery life, ensure safe operation of equipment, and reduce maintenance costs.

[0004] Traditional lithium-ion battery SOH evaluation methods can be divided into three categories: First, empirical methods evaluate the battery health state by constructing a mathematical formula between capacity degradation and operating conditions. Influencing factors include depth of discharge (DOD), temperature, rate (C-rate), storage time and coulombic efficiency. This type of method requires a large amount of data, and some parameters are highly nonlinear with SOH. Second, model methods rely on building accurate equivalent battery models to simulate battery behavior. These models include electrochemical models (EM) and equivalent circuit models (ECM). By combining algorithms such as Kalman filtering, the battery SO H is iteratively calculated. However, due to the continuous changes in actual working conditions, this method is limited in practical applications. Third, data-driven methods use machine learning algorithms to predict battery health status. Including support vector machines (SVM), Gaussian process regression (GPR), etc. However, these simple models are difficult to accurately capture the complex nonlinear process of battery degradation. Although deep learning methods have shown good performance, existing methods do not fully consider the common characteristics under different charging strategies.

[0005] In summary, the current SOH evaluation methods for lithium-ion batteries have problems such as inaccurate evaluation and limitations in practical applications. Summary of the invention

[0006] In order to overcome the shortcomings of the above-mentioned prior art, the present invention provides a lithium-ion battery health status assessment method and system. The extracted key features can effectively reflect the health status degradation performance of the battery, and the use of three different correlation methods for screening can effectively reduce feature noise, thereby improving the accuracy of subsequent evaluation.

[0007] In order to achieve the above object, the present invention adopts the following technical solution:

[0008] In a first aspect, the present invention provides a method for assessing the health status of a lithium-ion battery, comprising:

[0009] Obtaining the tangent slopes of different voltage points in the IC curve of the lithium-ion battery to be tested and the voltage values ​​corresponding to the extreme points, as well as the current characteristics of the constant current charge and discharge time and the voltage characteristics of the constant voltage charge time as the key characteristics of the lithium-ion battery to be tested;

[0010] The key features of the lithium-ion battery to be tested are screened by using the Pearson correlation coefficient, the Spearman rank correlation coefficient and the maximum information coefficient, respectively, to determine the health factor of the lithium-ion battery to be tested;

[0011] The health factor of the lithium-ion battery to be tested is input into an evaluation model to obtain an evaluation result of the lithium-ion battery to be tested, wherein the evaluation model is obtained by training a convolutional neural network and an extended long short-term memory network.

[0012] In a second aspect, the present invention provides a lithium-ion battery health status assessment system, comprising:

[0013] An acquisition unit is configured to: acquire the tangent slopes of different voltage points in the IC curve of the lithium ion battery to be tested and the voltage values ​​corresponding to the extreme points, as well as the current characteristics of the constant current charge and discharge time and the voltage characteristics of the constant voltage charge time, as key characteristics of the lithium ion battery to be tested;

[0014] A correlation calculation unit is configured to: use the Pearson correlation coefficient, the Spearman rank correlation coefficient and the maximum information coefficient to screen the key features of the lithium-ion battery to be tested, and determine the health factor of the lithium-ion battery to be tested;

[0015] An evaluation unit is configured to: input the health factor of the lithium-ion battery to be tested into an evaluation model to obtain an evaluation result of the lithium-ion battery to be tested, wherein the evaluation model is obtained by training a convolutional neural network and an extended long short-term memory network.

[0016] In a third aspect, the present invention provides an electronic device comprising a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method described in the first aspect is performed.

[0017] In a fourth aspect, the present invention provides a computer-readable storage medium for storing computer instructions, wherein when the computer instructions are executed by a processor, the method described in the first aspect is performed.

[0018] In a fifth aspect, the present invention provides a computer program product, comprising a computer program, which, when executed by a processor, implements the method described in the first aspect.

[0019] One or more of the above technical solutions have the following beneficial effects:

[0020] In the present invention, the tangent slopes of different voltage points in the discharge IC curve of the lithium-ion battery and the voltage values ​​corresponding to the extreme points, as well as the current characteristics of the constant current charge and discharge time and the voltage characteristics of the constant voltage charge time are obtained as the key characteristics of the lithium-ion battery to be tested, and the Pearson correlation coefficient, the Spearman rank correlation coefficient and the maximum information coefficient are used to perform correlation calculation to obtain the final health factor for evaluation; the key characteristics extracted by the present invention can effectively reflect the degradation performance of the health status of the battery, and the screening by three different correlation methods can effectively reduce the feature noise, thereby improving the accuracy of the subsequent evaluation.

[0021] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0023] Figure 1 This is a schematic diagram of the CNN-xLSTM network training process in Example 1 of the present invention;

[0024] Figure 2 Schematic diagram of the overall process of the lithium-ion battery health status assessment method in Example 1 of the present invention. DETAILED DESCRIPTION

[0025] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0026] It should be noted that the terms used herein are for describing specific embodiments only and are not intended to be limiting of exemplary embodiments according to the present invention.

[0027] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0028] Embodiment 1

[0029] This embodiment discloses a method for evaluating the health status of a lithium-ion battery, including:

[0030] Obtain the IC curve of the lithium-ion battery to be tested during the discharge process, and extract the tangent slope at different voltage points and the extreme points and voltage values ​​corresponding to the extreme points corresponding to the entire discharge IC curve, and at the same time extract the constant current charging and discharge duration as current characteristics and the constant voltage charging time as characteristics, and extract the above characteristics as key characteristics of the lithium-ion battery to be tested;

[0031] In order to achieve prediction accuracy, the extracted features of the lithium-ion battery to be tested are screened by using the Pearson correlation coefficient, Spearman correlation coefficient and maximum information coefficient respectively. By analyzing the correlation coefficients, the features with high correlation coefficients are screened as health factors. The health factors of the lithium-ion battery to be tested are input into the evaluation model to obtain the corresponding health status of the lithium-ion battery to be tested. The evaluation model is obtained by offline training of the convolutional neural network and the extended long short-term memory network (CNN-xLSTM) network.

[0032] This embodiment uses the sensor measurement parameters of the lithium-ion battery operating status to calculate: current curve, voltage curve and IC curve of the discharge process, extracts multidimensional related variables by analyzing the features at different locations of the curves, uses a combination of Pearson correlation coefficient, Spearman correlation coefficient and maximum information coefficient to screen the features extracted from the curves, selects the most relevant features as health factors, and achieves high-precision assessment of the health status by combining an improved three-layer extended long short-term memory network structure CNN-xLSTM with a convolutional neural network.

[0033] Combine the following Figure 1-Figure 2 A lithium-ion battery health status assessment method proposed in this embodiment is described in detail.

[0034] Step 1: extract the tangent slope at different voltage points in the IC curve of the lithium-ion battery to be tested during the discharge process, the extreme points of the entire discharge IC curve and their corresponding voltage values, as well as the current characteristics of the constant current charging and discharging time and the voltage characteristics of the constant voltage charging time as the key characteristics of the lithium-ion battery to be tested.

[0035] The degradation of the battery's health status is mainly manifested in capacity attenuation and internal resistance increase. These changes will be reflected in the specific voltage points and specific position characteristics of the incremental capacity (IC) curve, constant current charging and discharging in the voltage and current curve, and constant voltage charging process. When the battery's health status changes, the corresponding curve characteristics will also change regularly, so these characteristics can be used to evaluate the battery's health status.

[0036] In order to accurately capture the internal changes in the battery aging process, a variety of key features are extracted from the battery charge and discharge data, including the incremental capacity curve, that is, the IC curve of the discharge process, specific voltage points and specific position characteristics, constant current charge and discharge current characteristics, and constant voltage charge voltage characteristics. These features are of great significance in describing the changes in the battery health state. According to the changes in the curve characteristics, the health factor is extracted; the mapping relationship between the change in the health factor and the battery health state (SOH) is determined, and the battery health state is predicted using this mapping relationship. In particular, the incremental capacity (IC) curve characteristics can effectively reflect the internal degradation process of the battery under different charge and discharge strategies.

[0037] The innovation of this embodiment is that since lithium-ion batteries have different discharge voltage platforms, the tangent slope of the IC curve at a specific voltage point, the voltage value corresponding to the extreme point, etc. during the discharge process show similar change trends in different batteries and cycle stages. The area enclosed by dQ / dV=-1 and the IC curve can effectively represent the SOH changes of lithium-ion batteries under different discharge cycles, and these features can be used to make more accurate SOH predictions.

[0038] Specifically, the key features of the IC curve are: the tangent slope S0-S6 at a specific voltage point, the IC curve minimum S7 and the corresponding voltage S8, the tangent slope S9 at 0.05V before the extreme value, and the area S enclosed by dQ / dV=-1 and the IC curve. 10 . Key characteristics of current: 1C constant current charging time S 11 , 4C constant current discharge time S 12 . Key characteristics of voltage: constant voltage charging time S after 80% SOC 13 .

[0039] As an implementation method, the key features extracted from the IC curve include: the tangent slope value S0 at 3.3V, the tangent slope value S1 at 3.23V, the tangent slope value S2 at 3.18V, the tangent slope value S3 at 3.16V, the tangent slope value S4 at 3.13V, and the tangent slope value S5 at 2.9V; the slope value S6 of the straight line of the IC curve between 3.2V and 3.0V; the minimum value S7 of the IC curve and its corresponding voltage value S8; the tangent slope S9 at 0.05V before the minimum value; the area value S enclosed by the IC curve and the dQ / dV=-1 straight line 10 .

[0040] Key features extracted from the current curve: 1A constant current charging time S after 80% SOC 11 , 4A constant current discharge time S 12 .

[0041] The key features extracted from the voltage curve include: constant voltage charging time S13 after 80% SOC.

[0042] Step 2: Use the Pearson correlation coefficient, Spearman correlation coefficient and maximum information coefficient to screen the key features extracted from the lithium-ion battery to be tested. By analyzing the correlation coefficients, the features with high correlation coefficients are screened to determine the health factors of the lithium-ion battery to be tested.

[0043] In order to further improve the validity of features and the predictive ability of the evaluation model, the Pearson correlation coefficient, Spearman correlation coefficient and maximum information coefficient were used to screen out features extracted from the curve that were highly correlated with health status.

[0044] Feature selection based on Pearson correlation coefficient, the correlation coefficient calculation formula is:

[0045]

[0046] Among them, SOH i represents the actual value of the ith SOH, Represents the average value of SOH; S i represents the actual eigenvalue of the extracted feature, Represents the average value of this feature.

[0047] Feature selection based on Spearman correlation coefficient, the correlation coefficient calculation formula is:

[0048]

[0049] Among them, d i =rank(S i )-rank(SOH i )SOH i represents the actual value of the i-th SOH, S i

[0050] Represents the actual value of the i-th feature extracted.

[0051] Feature selection based on the maximum information coefficient is calculated as follows:

[0052]

[0053] in:

[0054]

[0055] Among them, SOH represents the actual value of the SOHth element; S represents the actual value of a certain feature extracted.

[0056] The extremely strong correlation features with correlation coefficients of the three groups all greater than a set threshold value such as 0.8 are extracted as the health factors of the ion battery to be tested.

[0057] Step 3: Input the health factor of the lithium-ion battery to be tested into the trained evaluation model to obtain the corresponding health status of the lithium-ion battery to be tested. The evaluation model is obtained by offline training of a convolutional neural network and an extended long short-term memory network (CNN-xLSTM).

[0058] CNN-xLSTM network model: A multi-scale convolutional neural network (CNN) structure is introduced before the xLSTM network. The multi-scale convolutional neural network is used to extract local spatial features of battery data. The xLSTM network is composed of sLSTM and mLSTM of different layers; different sLSTM and mLSTM are connected through a residual network. The xLSTM network used in this embodiment adopts an optimized three-layer [s, m, s] network architecture.

[0059] The structures of the first and third layers are sLSTM. sLSTM extends LSTM by introducing layer normalization and causal convolution to enhance the ability to capture time series features. In terms of gating mechanism, sLSTM implements dynamic gating by introducing max operation and exponential function, replacing the simple sigmoid gating in traditional LSTM, so that the model can control the flow of information more flexibly and in terms of feature extraction; sLSTM introduces an additional normalization term n t , which is used to regulate the update of hidden states. This design improves the numerical stability of the model.

[0060] The first layer of sLSTM in this embodiment is used for preliminary feature extraction and processing. The second layer uses the mLSTM structure to implement a matrix memory mechanism to improve feature extraction capabilities, and introduces two 256-dimensional attention heads to handle sequence dependencies. The first attention mechanism: Key-ValueAttention is used to update the cell state. The second attention mechanism: Query-based Attention, this attention mechanism is used to generate the final hidden state. The third layer of sLSTM integrates the feature information extracted by the first two layers, and the final feature projection prepares the final output. This innovative network structure combined with CNN can better capture the complex sequence features in the battery degradation process and achieve accurate prediction of SOH. Residual connections are used between the layers of xLSTM; residual connections are also used between CNN and the entire xLSTM to transmit feature information.

[0061] Specifically, CNN extracts the local spatial features of battery data, and the formula is expressed as:

[0062]

[0063] Among them, S t represents the input feature vector at time t, w i is the weight parameter of the convolution kernel, and b is the bias term.

[0064] Adding residual connections to CNN enables direct transmission of original information, effectively retaining important basic features, while also alleviating the gradient vanishing problem in deep network training.

[0065]

[0066] The superscript s indicates short-term scale extraction. t represents the input feature vector at time t, w i is the weight parameter of the convolution kernel, and b is the bias term.

[0067]

[0068] The superscript m indicates that the short-term scale extraction S t represents the input feature vector at time t, w i is the weight parameter of the convolution kernel, and b is the bias term.

[0069] As an implementable method, CNN extracts features at two scales, medium-term and short-term: convolution layers at two scales, short-term and medium-term, are designed to capture the local features and trend features of the battery health status over time. Among them, the short-term feature convolution uses a 3×1 convolution kernel to extract the short-term fluctuation features of the battery parameters; the medium-term feature convolution uses a 5×1 convolution kernel to capture the attenuation trend of the battery capacity.

[0070] The first layer sLSTM and the third layer sLSTM introduce an adaptive gating mechanism, which dynamically adjusts the weights of the forget gate and the input gate to enhance the robustness to noise, adds residual connections to alleviate the gradient vanishing problem, and uses layer normalization technology to improve training stability:

[0071] c t =f t ⊙c t-1 +i t ⊙z t (8)

[0072] n t =f t ⊙n t-1 +i t (9)

[0073]

[0074] Among them, for sLSTM, c t is the cell state, n t is the sLSTM normalized state, z t is a candidate memory state, used to update the cell state c t ;h t is the hidden state. Input gate i t 、Forget gate t and output gate o t is defined as follows:

[0075]

[0076] Among them, x t is the input vector of sLSTM at time step t, W and R are weight matrices, and b i : Bias term of input gate (InputGate). b f : is the bias term of the forget gate (ForgetGate). b o : The bias term of the output gate (OutputGate); the activation functions σ and exp represent the sigmoid function and the exponential function respectively.

[0077] To prevent possible overflows caused by the exponential function, an additional stabilizer state m is used. t To stabilize the gate:

[0078] m t =max(log(f t )+m t-1 ,log(i t )) (14)

[0079] i t′ =exp(log(i t )-m t ) (15)

[0080] f t′ =exp(log(f t )+m t-1 -m t ) (16)

[0081] By using f in formula (16) in the forward propagation t′ Substituting f in formula (12) t , using formula (15) t′ Substitute i in formula (11) t , sLSTM maintains stability without changing the loss gradients with respect to outputs or parameters.

[0082] The middle layer mLSTM is the matrix memory layer: it expands the memory unit from scalar to matrix form, significantly improves the information storage capacity, designs a new attention mechanism, realizes efficient information retrieval through the attention mechanism query-key-value, introduces an adaptive weight update strategy, and dynamically adjusts the update rate of the memory matrix:

[0083]

[0084] nt=ftnt-1+itkt (18)

[0085]

[0086] v t =W v x t +b v (twenty one)

[0087] q t =W q x t +b q (twenty two)

[0088] For mLSTM, C t is the matrix memory state, n t is the normalized state of mLSTM, h t is the hidden state. The keyword vector k t , numerical vector v t and the query vector q t , x t is the input vector of mLSTM at time step t, W represents the weight matrix, b k The bias term when generating the keyword vector (Key). v The bias term when the numerical vector (Value) is generated. q Bias term when generating query vector (Query).

[0089]

[0090] Among them, x t is the input vector at time step t, W represents the weight matrix, and b is the bias term. The activation functions σ and exp represent the sigmoid function and exponential function respectively.

[0091] In order to stabilize the exponential gating and prevent overflow, the same stabilization technique as sLSTM is used, introducing a stabilizer state mt.

[0092] m t =max(log(f t )+m t-1 ,log(it )) (26)

[0093] i t′ =exp(log(i t )-m t ) (27)

[0094] f t′ =exp(log(f t )+m t-1 -m t ) (28)

[0095] By using f in formula (28) in the forward propagation t′ Substituting f in formula (24) t , using formula (27) t′ Substitute i in formula (23) t , mLSTM maintains stability without changing the loss gradients with respect to the output or parameters.

[0096] The third layer, sLSTM, is the comprehensive layer: it considers both historical and future information, integrates the feature information extracted by the first two layers, and prepares the final output through the final feature projection. It integrates feature representations at different levels and dynamically adjusts weights based on the feature information input from the first two layers.

[0097] CNN is used to extract the short-term and medium-term feature scales of selected health factors; combined with the attention mechanism of LSTM; the importance of features is analyzed from different scales, thereby achieving accurate SOH prediction. Three different extended LSTM layers are connected through a residual network:

[0098] xLSTM={L1,L2,...,L n}, L i ∈{s,mLSTM} (29)

[0099] Among them, L i Represents different,s,mLSTM layers.

[0100] This embodiment uses 124 commercial lithium iron phosphate (LFP) batteries in the MIT data set as target batteries, with a rated capacity of 1.1Ah and a nominal voltage of 3.3V. These batteries were cycled in a 48-channel ArbinLBT battery test system at a constant temperature of 30°C, including 72 different charge and discharge strategies. Based on the internal degradation mechanism of the battery during the cycle, the extracted features include specific voltage points and specific position features of the discharge process (IC) curve, current characteristics of constant current charge and discharge time, and voltage characteristics of constant voltage charging time. Three correlation coefficients are used for analysis, and features with high correlation coefficients are screened out as health factors. The obtained health factors are used as training samples, and the proposed CNN-xLSTM network is trained using the training samples to obtain a trained CNN-xLSTM network as an evaluation model.

[0101] The training process of the convolutional neural network and the CNN-xLSTM network mainly includes the following steps: Standardize the input health factors:

[0102]

[0103] Where X is the original data, Xmin and Xmax are the minimum and maximum values ​​of X respectively.

[0104] The processed data is input into the convolutional neural network and xLSTM network, and the predicted SOH value is calculated through forward propagation; then, the error between the predicted value and the true value is calculated using the loss function MSE; then, the error is passed back to each layer of neurons using the back propagation algorithm, and the weights and biases of the model are gradually adjusted to minimize the loss function. The feature transfer between layers is:

[0105]

[0106] in, are the hidden state and unit state of the i-th layer at time step t, respectively.

[0107] In this embodiment, (1) batch training is adopted, and the batch size is set to 256. The weight and bias parameters of the model are initialized, and the first batch of 256 groups of samples are simultaneously input into the network. The prediction results of the current batch are calculated through forward propagation, and the loss function value is calculated;

[0108] (2) Update the model parameters through back propagation based on the calculated loss function value. The loss function includes the mean square error loss:

[0109]

[0110] Among them, y i Indicates the true SOH value, represents the predicted SOH value, and n represents the number of samples.

[0111] (3) When training the next batch, the model parameters use the parameter values ​​after the previous batch training, continue to update the parameters through forward propagation and back propagation, obtain the average loss of the current batch, and update the model parameters accordingly.

[0112] (4) Repeat steps (1)-(3) until the preset number of training rounds of 30 or the loss function converges. The model uses the Adam optimizer, and the initial learning rate is set to 0.0001. Ensure efficient convergence during the training process and prevent the model from overfitting, thereby improving the generalization ability of the model.

[0113] The test set battery data is input into the trained model for testing to evaluate the prediction accuracy. In this embodiment, eight batteries with different charge and discharge strategies are selected as the test set, and the mean absolute percentage error (MAPE), root mean square error (RMSE) and determination coefficient (R 2 )Three indicators are used to evaluate the model performance.

[0114] The CNN-xLSTM network learns the battery aging pattern through data under different charging and discharging strategies. The CNN-xLSTM network training process adopts supervised learning and optimizes the model using labeled battery health status (SOH) data.

[0115] The innovation of CNN-xLSTM lies in the use of residuals to connect different layers of the network; CNN performs multi-scale feature combination extraction, captures feature dependencies in the input sequence through the first sLSTM layer, introduces multiplicative interactions through the mLSTM layer, and uses the matrix memory unit of mLSTM to make the model more adaptable in the face of complex industrial scenarios, thereby better assessing the health status.

[0116] This embodiment constructs an evaluation model that can accurately describe the battery health status through the mapping relationship between the specific voltage points and specific position characteristics of the IC curve during the discharge process, the constant current charge and discharge current characteristics and the battery SOH. The CNN-xLSTM network is used to perform deep learning on the features, and the long-term and short-term dependency features are extracted through CNN. The feature capture of the sLSTM layer and the matrix memory mechanism of the mLSTM layer effectively capture the complex sequence dependency in the battery degradation process, and achieve high-precision prediction of the battery health status.

[0117] Embodiment 2

[0118] The purpose of this embodiment is to provide a lithium-ion battery health status assessment system, including:

[0119] An acquisition unit is configured to: acquire the tangent slopes of different voltage points in the IC curve of the lithium ion battery to be tested during the discharge process and the voltage values ​​corresponding to the extreme points, as well as the constant current charge and discharge current characteristics and the constant voltage charge voltage characteristics, as key characteristics of the lithium ion battery to be tested;

[0120] A correlation calculation unit is configured to: use the Pearson correlation coefficient, the Spearman rank correlation coefficient and the maximum information coefficient to screen the key features of the lithium-ion battery to be tested, and determine the health factor of the lithium-ion battery to be tested;

[0121] An evaluation unit is configured to: input the health factor of the lithium-ion battery to be tested into an evaluation model to obtain an evaluation result of the lithium-ion battery to be tested, wherein the evaluation model is obtained by training a CNN-xLSTM network.

[0122] In further embodiments, there is also provided:

[0123] An electronic device includes a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method described in Embodiment 1 is performed. For the sake of brevity, it will not be described in detail here.

[0124] It should be understood that in this embodiment, the processor may be a central processing unit CPU, and the processor may also be other general-purpose processors, digital signal processors DSP, application-specific integrated circuits ASIC, off-the-shelf programmable gate arrays FPGA 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.

[0125] The memory may include a read-only memory and a random access memory, and provide instructions and data to the processor. A portion of the memory may also include a non-volatile random access memory. For example, the memory may also store information about the device type.

[0126] A computer-readable storage medium is used to store computer instructions. When the computer instructions are executed by a processor, the method described in embodiment 1 is completed.

[0127] The method in the first embodiment can be directly embodied as a hardware processor, or a combination of hardware and software modules in the processor. The software module can be located in a mature storage medium in the field such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory, and the processor reads the information in the memory and completes the steps of the above method in combination with its hardware. To avoid repetition, it will not be described in detail here.

[0128] A computer program product includes a computer program, and when the computer program is executed by a processor, the method described in the first embodiment is implemented.

[0129] The present invention also provides at least one computer program product tangibly stored on a non-transitory computer-readable storage medium. The computer program product includes computer executable instructions, such as instructions included in a program module, which are executed in a device on a real or virtual processor of the target to perform the process / method as described above. Typically, program modules include routines, programs, libraries, objects, classes, components, data structures, etc. that perform specific tasks or implement specific abstract data types. In various embodiments, the functions of program modules can be combined or divided between program modules as needed. Machine executable instructions for program modules can be executed in local or distributed devices. In distributed devices, program modules can be located in local and remote storage media.

[0130] The computer program code for implementing the method of the present invention can be written in one or more programming languages. These computer program codes can be provided to the processor of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the computer or other programmable data processing device, causes the function / operation specified in the flow chart and / or block diagram to be implemented. The program code can be executed completely on a computer, partially on a computer, as an independent software package, partially on a computer and partially on a remote computer or completely on a remote computer or server.

[0131] In the context of the present invention, computer program codes or related data may be carried by any appropriate carrier to enable a device, apparatus or processor to perform the various processes and operations described above. Examples of carriers include signals, computer readable media, etc. Examples of signals may include electrical, optical, radio, acoustic or other forms of propagation signals, such as carrier waves, infrared signals, etc.

[0132] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with this embodiment can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0133] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A method for evaluating the health status of a lithium-ion battery, characterized in that: include: Obtaining the tangent slopes of different voltage points in the IC curve of the lithium-ion battery to be tested during the discharge process and the voltage values ​​corresponding to the extreme points, as well as the current characteristics of the constant current charge and discharge time and the voltage characteristics of the constant voltage charge time as the key characteristics of the lithium-ion battery to be tested; The key features of the lithium-ion battery to be tested are screened by using the Pearson correlation coefficient, the Spearman rank correlation coefficient and the maximum information coefficient, respectively, to determine the health factor of the lithium-ion battery to be tested; The health factor of the lithium-ion battery to be tested is input into an evaluation model to obtain an evaluation result of the lithium-ion battery to be tested, wherein the evaluation model is obtained by training a CNN-xLSTM network.

2. A lithium-ion battery health status assessment method as claimed in claim 1, characterized in that: The constant current charging and discharging current characteristics include: 1C constant current charging time and 4C constant current discharging time; the constant voltage charging voltage characteristics include: constant voltage charging time after 80% SOC.

3. A lithium-ion battery health status assessment method as claimed in claim 1, characterized in that: The key features of the lithium ion battery to be tested are screened using the Pearson correlation coefficient, the Spearman rank correlation coefficient and the maximum information coefficient, respectively, to determine the health factor of the lithium ion battery to be tested, specifically: the key features corresponding to the correlation coefficients calculated using the Pearson correlation coefficient, the Spearman rank correlation coefficient and the maximum information coefficient are greater than the set threshold value, respectively, as the determined health factor of the lithium ion battery to be tested.

4. A lithium-ion battery health status assessment method as claimed in claim 1, characterized in that: The process of processing the health factor of the lithium-ion battery to be tested by the evaluation model is as follows: Using CNN to collect local spatial characteristics of the health factor of the lithium-ion battery to be tested; The sLSTM network layer, the mLSTM network layer and the sLSTM network layer are used to capture the complex time dependency and nonlinear relationship between the local spatial features of the input and the battery health state, and obtain the evaluation result of the lithium-ion battery to be tested.

5. A lithium-ion battery health status assessment method as claimed in claim 4, characterized in that: The method of using CNN to collect the local spatial features of the health factor of the lithium-ion battery to be tested is specifically as follows: using convolutional layers of different scales to respectively extract the short-term features and medium-term features of the health assessment factor of the lithium-ion battery to be tested, and fusing the extracted short-term features and medium-term features to obtain local spatial features.

6. A lithium-ion battery health status assessment method as claimed in claim 4, characterized in that: An attention mechanism is introduced into the mLSTM network layer to dynamically allocate weight coefficients according to the importance of different scale features to SOH prediction, thereby realizing adaptive combination of features.

7. A lithium-ion battery health status assessment system, characterized in that: include: An acquisition unit is configured to: acquire the tangent slopes of different voltage points in the IC curve of the lithium ion battery to be tested during the discharge process and the voltage values ​​corresponding to the extreme points, as well as the constant current charge and discharge current characteristics and the constant voltage charge voltage characteristics, as key characteristics of the lithium ion battery to be tested; A correlation calculation unit is configured to: use the Pearson correlation coefficient, the Spearman rank correlation coefficient and the maximum information coefficient to screen the key features of the lithium-ion battery to be tested, and determine the health factor of the lithium-ion battery to be tested; An evaluation unit is configured to: input the health factor of the lithium-ion battery to be tested into an evaluation model to obtain an evaluation result of the lithium-ion battery to be tested, wherein the evaluation model is obtained by training a CNN-xLSTM network.

8. An electronic device, characterized in that: The method comprises a memory and a processor and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the method according to any one of claims 1 to 6 is completed.

9. A computer-readable storage medium, characterized in that: Used to store computer instructions, which, when executed by a processor, complete the method described in any one of claims 1 to 6.

10. A computer program product, characterized in that The invention comprises a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 6.

Images