Lithium ion battery residual life prediction method based on SABO-ELM model and related device

Abstract

The invention discloses a lithium ion battery remaining life prediction method based on an SABO-ELM model and a related device, and belongs to the technical field of battery life prediction.The method comprises the steps that an electric quantity increment curve of a lithium ion battery is obtained, and health factors are extracted through the electric quantity increment curve; after the correlation strength between the health factors and the capacity of the lithium ion battery is evaluated, remarkably related health factors are screened out; and after optimizing a weight and a bias threshold value of an extreme learning machine by adopting a subtraction average algorithm, inputting the significant correlation health factor into the extreme learning machine for training to obtain an SABO-ELM model, and predicting the residual life of the lithium ion battery through the SABO-ELM model. According to the method, the RUL prediction precision of the lithium ion battery can be improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of battery life prediction, and in particular relates to a method and a related device for predicting the remaining life of a lithium-ion battery based on a SABO-ELM model. Background Art

[0002] Lithium-ion batteries have shown great potential in solving energy depletion and environmental crises. Due to their advantages such as high energy density, light weight and long cycle time, lithium-ion batteries have been widely used in electric vehicles, aerospace and consumer electronics. As the number of charge and discharge cycles of lithium-ion batteries increases during long-term operation, some complex reactions will occur inside them, causing irreversible aging. This not only affects the use of related electronic equipment, but also brings some potential risks. Therefore, in practical applications, it is very important to accurately predict the remaining useful life (RUL) of lithium-ion batteries in advance.

[0003] At present, the remaining life estimation methods are mainly divided into two types: model-based and data-driven. Model-based methods are divided into equivalent circuit models and electrochemical models according to different modeling mechanisms. The electrochemical model is based on the physical and chemical reactions inside the battery, combined with the characteristics of different electrode materials, to derive a complete mathematical model to describe the law of battery degradation. Therefore, this method is to achieve RUL prediction by analyzing the failure mechanism and electrochemical reaction of the battery. However, the model-driven method requires an accurate reference model and appropriate parameter settings, and needs to analyze the complex internal mechanism and electrochemical reaction process of the battery. The modeling process is complex and the environmental factors interfere greatly. Therefore, the data-driven method eliminates the need to build a specific prediction model based on the complex law of battery degradation, and uses charging and discharging data to fully train machine learning or deep learning models. Data-driven methods include evolutionary algorithms, machine learning algorithms, neural network algorithms, etc. However, the current data-driven methods do not make more reasonable selection and optimization of the model when predicting RUL, which affects the accuracy of the RUL prediction results. Summary of the invention

[0004] The purpose of the present invention is to provide a method and related device for predicting the remaining service life of a lithium-ion battery based on the SABO-ELM model, so as to improve the accuracy of the RUL prediction of lithium-ion batteries in the prior art.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] In a first aspect, a method for predicting the remaining life of a lithium-ion battery based on a SABO-ELM model comprises the following steps:

[0007] Obtaining a charge increment curve of the lithium-ion battery, and extracting a health factor through the charge increment curve;

[0008] After evaluating the strength of the correlation between the health factor and the capacity of the lithium-ion battery, a significantly correlated health factor is screened out;

[0009] After optimizing the weights and bias thresholds of the extreme learning machine by using the subtraction mean algorithm, the significantly correlated health factors are input into the extreme learning machine for training to obtain a SABO-ELM model, and the remaining life of the lithium-ion battery is predicted by the SABO-ELM model.

[0010] In some embodiments, after collecting the capacity differential curve and the voltage differential curve during the constant current charging process of the lithium ion battery, the charge increment curve is obtained according to the capacity differential curve and the voltage differential curve.

[0011] In some embodiments, the health factor includes: a dQ / dV peak value, a dQ / dV peak position, and a dQ / dV value other than the dQ / dV peak value.

[0012] In some embodiments, the Spearman method is used to evaluate the strength of the correlation between the health factor and the capacity of the lithium-ion battery.

[0013] In some implementations, the step of optimizing the weights and bias thresholds of the extreme learning machine using a subtraction mean algorithm specifically includes:

[0014] After setting the population size and the number of iterations and assigning the initial positions of the search agents, the objective function value of the initial positions is calculated;

[0015] The suggested position of the search agent is calculated by using a subtraction mean algorithm, and the objective function value of the suggested position is calculated. If the objective function value of the suggested position is better than the objective function value of the initial position, the initial position of the search agent is updated to the suggested position, otherwise it is not updated, and the above steps are repeated until the number of iterations is met, and the suggested position of the search agent that is finally updated is output;

[0016] The population size is the number of search agents, and the positions of the search agents are the weights and bias thresholds of the extreme learning machine.

[0017] In some implementations, in the step of extracting the health factor through the power increment curve, the power increment curve is first subjected to Kalman filtering, and then the health factor is extracted from the power increment curve.

[0018] In the second aspect, a lithium-ion battery remaining life prediction system based on the SABO-ELM model comprises:

[0019] A health factor extraction module is used to obtain a power increment curve of the lithium-ion battery, and extract the health factor through the power increment curve;

[0020] A significant health factor screening module is used to screen out significant related health factors after evaluating the correlation strength between the health factors and the capacity of the lithium-ion battery;

[0021] The extreme learning machine optimization prediction module is used to optimize the weights and bias thresholds of the extreme learning machine by using a subtraction mean algorithm, input the significantly correlated health factors into the extreme learning machine for training to obtain a SABO-ELM model, and predict the remaining life of the lithium-ion battery by using the SABO-ELM model.

[0022] In a third aspect, an electronic device includes a memory, a processor, and a computer program stored in the memory and executable in the processor, wherein the processor implements the steps of the method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model when executing the computer program.

[0023] In a fourth aspect, a computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model are implemented.

[0024] In a fifth aspect, a computer program product is provided, the computer product comprising a computer program, characterized in that when the computer program is executed by a processor, the steps of the method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model are implemented.

[0025] Compared with the prior art, the present invention has the following beneficial effects:

[0026] The present invention provides a method for predicting the remaining service life of a lithium-ion battery based on a SABO-ELM model. The method comprises the following steps: firstly obtaining a charge increment curve of a lithium-ion battery, and extracting a health factor through the charge increment curve; after evaluating the correlation strength between the health factor and the capacity of the lithium-ion battery, screening out a significantly correlated health factor; after optimizing the weight and bias threshold of an extreme learning machine by using a subtraction average algorithm, inputting the significantly correlated health factor into the extreme learning machine for training to obtain a SABO-ELM model; and predicting the remaining service life of a lithium-ion battery through the SABO-ELM model, which can further improve the prediction capability of the remaining useful life (RUL) of the battery. Moreover, by comparing with other models in the embodiments, the SABO-ELM model provided by the present invention has higher accuracy for RUL prediction, and its maximum root mean square error does not exceed 2%, and its maximum absolute error does not exceed 1.5%.

[0027] Furthermore, the present invention preliminarily selects dQ / dV peak values, dQ / dV peak positions and dQ / dV values ​​under some peak intervals in the incremental capacity (IC) curve as health factors (HF), and then uses Spearman to analyze the strength of the correlation between HF and the capacity of the lithium-ion battery, and screens out significantly correlated HF with strong correlation, which can solve the current problem of inaccurate HF direct measurement and feature extraction. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 The overall framework diagram of the method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model provided in Example 1;

[0029] Figure 2 The battery capacity attenuation diagram of the batteries B05, B06 and B07 used in Example 1;

[0030] Figure 3 It is a comparison diagram of the IC curve before and after filtering in Example 1;

[0031] Figure 4 This is a health factor graph extracted from the IC curve in Example 1;

[0032] FIG5( a ) is a correlation analysis diagram of the health factor of battery B05 in Example 1;

[0033] FIG5( b ) is a correlation analysis diagram of the health factor of battery B06 in Example 1;

[0034] FIG5( c ) is a correlation analysis diagram of the health factor of battery B07 in Example 1;

[0035] Figure 6 This is a schematic diagram of the process of using SABO to optimize ELM in Example 1;

[0036] FIG. 7( a ) is a diagram showing the RUL prediction results of battery B05 at different prediction starting points in Example 1;

[0037] FIG. 7( b ) is a relative error diagram of the RUL prediction results of the B05 battery at different prediction starting points in Example 1;

[0038] FIG. 7( c ) is a diagram showing the RUL prediction results of battery B06 at different prediction starting points in Example 1;

[0039] FIG. 7( d ) is a relative error diagram of the RUL prediction results of the B06 battery at different prediction starting points in Example 1;

[0040] FIG. 7( e ) is a diagram showing the RUL prediction results of battery B07 at different prediction starting points in Example 1;

[0041] FIG. 7( f ) is a relative error diagram of the RUL prediction results of the B07 battery at different prediction starting points in Example 1;

[0042] Figure 8 This is a schematic diagram of the structure of a lithium-ion battery remaining life prediction system based on the SABO-ELM model provided in Example 2. DETAILED DESCRIPTION

[0043] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solution of the present invention will be further described in detail below in conjunction with the accompanying drawings, and the described content is intended to explain the present invention rather than to limit it.

[0044] It should be noted that the terms "including" and "having" and any variations thereof in the specification and claims of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or are inherent to these processes, methods, systems, products or apparatus.

[0045] Embodiment 1

[0046] This embodiment provides a method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model. The experimental data of lithium-ion batteries used is a public data set of the National Aeronautics and Space Administration (NASA), and the batteries numbered B05, B06, and B07 are used as experimental objects. A 18650 lithium-ion battery with a nominal capacity of 2Ah is used. First, charge at a constant current of 1.5A until the terminal voltage reaches 4.2V, ending the constant current stage and entering the constant voltage stage. The battery continues to be charged in the constant voltage stage until the battery current drops from 1.5A to 20mA. In the discharge stage, a constant current of 2.0A is used to discharge to the cut-off voltage. The cut-off voltages of B05, B06, and B07 batteries are 2.7V, 2.5V, and 2.2V, respectively. In the repeated charge and discharge process, the failure threshold of the lithium-ion battery is defined as 70% of the capacity (1.4Ah). Since B07 has not dropped to 1.4Ah, 1.45Ah is used as its failure threshold. The experiment is terminated when the failure threshold is reached. Figure 2 Shown is the capacity decay curve of each battery;

[0047] The method for predicting the remaining life of a lithium-ion battery specifically comprises the following steps:

[0048] S1, obtaining a charge increment curve of a lithium-ion battery, and extracting a health factor through the charge increment curve;

[0049] The IC curve is obtained by taking the first-order derivative of the battery voltage-capacity curve during the battery charging and discharging process, which is specifically obtained by the following formula (1):

[0050]

[0051] Wherein, Q represents the current capacity of the battery, V represents the current terminal voltage value of the battery, I represents the current in the discharge phase, and t represents the time in the discharge phase. Since the IC curve has noise, this embodiment uses Kalman filtering to smooth the IC curve, such as Figure 3 As shown in the figure, after noise reduction processing, the IC curve is smoother and the peak is more obvious.

[0052] After the noise reduction process, the HF in the IC curve is extracted. The conventional method of selecting the HF in the IC curve mostly uses the area between the peak and the valley. This extraction method does not consider the change of the peak in the curve. In actual application, the charging process is generally not carried out after the battery is fully consumed. The charging range is generally within the range of 40% to 90%.

[0053] Therefore, in this embodiment, the peak interval (ΔV reg ). Use fixed peak width to extract other IC characteristic parameters. ΔV reg The voltages at the left and right ends of the defined peak region are:

[0054]

[0055] Where V peak is the peak voltage; V1(t1) is ΔV reg The voltage on the left side at time t1; V2(t2) is ΔV reg At time t2, the voltage on the right side is, therefore, based on the analysis, 3.85V to 4.09V is used as the voltage selection range, 0.03V is used as the selection interval, a dQ / dV value is taken every 0.03V, and they are used as features HF1 to HF8. The peak value and the position of the peak value are used as HF9-HF10, a total of 10 HFs, such as Figure 4 shown.

[0056] S2, after evaluating the correlation strength between the health factor and the lithium-ion battery capacity, screening out significantly correlated health factors;

[0057] In order to verify the correlation between the selected HF and capacity, the most representative significantly correlated HF was selected by eliminating redundant information, and the Spearman correlation analysis method was used to perform correlation analysis on the selected 10 HFs. The Spearman rank correlation coefficient was recorded as ρ, and the range was [-1,1]. ρ=-1 is completely negatively correlated, ρ=0 is no correlation, and ρ=1 is completely positively correlated. 0.89≤ρ<1 is extremely strongly correlated, 0.6<ρ<0.89 is strongly correlated, and ρ<0.6 is moderately correlated or weakly correlated.

[0058] The strength of the correlation between the screened HF and capacity is shown in Figure 5(a), Figure 5(b), and Figure 5(c). The six HFs screened for B05 and B07 batteries are (HF1-5, HF9), and the seven HFs for B06 batteries are (HF1-5, HF9, HF10). There is a strong correlation between them and the capacity, which can effectively express the battery capacity degradation trend.

[0059] S3, after optimizing the weights and bias thresholds of the extreme learning machine by using a subtraction mean algorithm, the significantly correlated health factors are input into the extreme learning machine for training to obtain a SABO-ELM model, and the remaining life of the lithium-ion battery is predicted by the SABO-ELM model.

[0060] In the Extreme Learning Machine (ELM), the weights connecting the input layer and the hidden layer and the bias of the hidden layer are randomly generated. Learning can be done more efficiently without iterative adjustments, and the weights connecting the hidden layer and the output layer are determined by fitting the training data.

[0061] The ELM network consists of an input layer, a hidden layer, and an output layer. The input layer contains n neurons, corresponding to n input variables; the hidden layer contains j neurons; and the output layer contains m neurons, corresponding to m output variables. Assume that the connection weight w between the input layer and the hidden layer is:

[0062]

[0063] Assume the connection weight β between the hidden layer and the output layer is:

[0064]

[0065] where β i is the output weight matrix. Let the threshold b of the hidden layer neuron be:

[0066]

[0067] Assume that the capacity of the battery sample can be expressed as (X i ,t i), where X i =[x i1 ,x i2 ,…,x in ] T ∈R n t i =[t i1 ,t i2 ,…,t im ] T ∈R m , for a single hidden layer extreme learning machine network with L constructs:

[0068]

[0069] Among them, g(X) is the activation function, W i is the input weight matrix.

[0070] In order to minimize the error:

[0071]

[0072] That is, there is β i , W and b i , so that:

[0073]

[0074] The matrix can be represented as:

[0075] Hβ=T(10)

[0076] Among them, H is the output of the hidden layer node, T is the expected output, and in order to obtain a suitable β to minimize the error function value or close to 0, we make:

[0077]

[0078] The solution is:

[0079]

[0080] Among them, H + is the generalized inverse of the hidden layer output matrix H.

[0081] In the Subtraction-Average-Based Optimizer (SABO), each optimization problem has a solution space, called the search space. The search space is a subset of the dimensional space that is equal to the number of decision variables for a given problem. Based on their position in the search space, the algorithm identifies agents (i.e., group members) that determine the values ​​of the decision variables. Therefore, each search agent contains information about the decision variables and is mathematically modeled using a vector

[26] The collection of search agents together forms the population of the algorithm. From a mathematical point of view, according to equation (13), the population of the algorithm can be represented using a matrix. The primary positions of the search agents in the search space are randomly initialized using equation (14).

[0082]

[0083] Where X is the SABO population matrix, X i is the i-th group member, x i,d is its dth dimension in the search space, is the number of search agents, m is the number of decision variables, N r i,d is a random number in the interval [0,1], lb d andub d are the lower and upper bounds of the dth decision variable, respectively.

[0084] SABO's concept of calculating the arithmetic mean is completely unique, as it is based on a special operation "-v", called v-subtraction of search agent B from search agent A, which is defined as follows:

[0085]

[0086] in is a randomly generated vector from the set {1,2}, and F(A) and F(B) are the values ​​of the objective function of the search agents A and B. By selecting i Subtract each search agent X j ,j=1,2,...,N,to calculate the displacement of any search agent Xi in the search space. Therefore, the new position of each search agent is calculated using (16).

[0087]

[0088] in Is the i-th search agent X i is the new suggested location of , N is the total number of search agents, is a vector of dimension m, where the components have a normal distribution with values ​​from the interval [0,1]. Then according to formula (17), if the proposed new position leads to an improvement in the value of the objective function, it can be accepted as the new position of the corresponding agent.

[0089]

[0090] where F i and They are search agent X i and The objective function value of .

[0091] Based on the above content, the flowchart of this embodiment using SABO to optimize the ELM model is as follows: Figure 6 shown.

[0092] In order to verify the prediction effect of the SABO-ELM model on the remaining life of lithium-ion batteries, this example uses the following three statistics: root mean square error RMSE, mean absolute error MAE, and mean absolute percentage error MAPE to evaluate the prediction results. The corresponding calculation formula is as follows:

[0093]

[0094] Among them, y i represents the true value of the i-th capacity of the battery, is an estimated value, and N is the number of battery cycles.

[0095] like Figure 7(a) to Figure 7(f) As shown in the figure, in order to verify the effectiveness and accuracy of the proposed method at different stages, the screened HF was used as the data set, and 70%, 60%, and 50% of them were used as training sets for training, respectively, and the remaining data was used as the test set and input into the SABO-ELM model to obtain the RUL prediction value.

[0096] Figure 7(a) to Figure 7(f) The RUL prediction results of NASA batteries B05, B06, and B07 at different prediction starting points are shown, where the brown-green solid line in the figure is the line where the error between the actual capacity and the predicted capacity is 0. As the prediction starting point increases, the predicted RUL result is closer to the actual RUL value. As shown in Table 1 below, when the training set is small, the model training is not sufficient and cannot be fully learned. The model error increases but the maximum error does not exceed 2.2%. This shows that the RUL method proposed in this embodiment is effective for predicting the RUL of lithium-ion batteries.

[0097] Table 1 RUL estimation error at different prediction starting points

[0098]

[0099] To further demonstrate the effectiveness of the proposed model, the RUL prediction accuracy is compared using ELM, SABO-ELM, Beluga whale optimization (BWO)-ELM, and LSTM models on the three types of batteries mentioned above by NASA, as shown in Table 2.

[0100] Table 2 RUL estimation errors of different models

[0101]

[0102] It is obtained in Table 2 that the estimation error of the SABO-ELM model established in this embodiment is the smallest. Taking the B07 battery as an example, the MAE, MAPE, and RMSE of the model proposed in this embodiment are 1.42%, 0.97%, and 1.91%, respectively, while the corresponding errors of the ELM model are 5.36%, 4.8%, and 3.26%, respectively; the corresponding errors of the ELM model (BWO-ELM) after optimization by the White Whale algorithm are 1.5%, 1.03%, and 2.03%; in addition, compared with the widely used LSTM model, the SABO-ELM model proposed in this embodiment reduces the prediction errors by 1.54%, 1.05%, and 1.44%, respectively. Among the three batteries, the maximum root mean square error of the SABO-ELM model is 1.96%, and the maximum mean absolute error and the maximum mean absolute percentage error are 1.42% and 1.01%. It further proves the applicability and accuracy of the remaining life prediction method of lithium-ion batteries based on the SABO-ELM model proposed in this embodiment.

[0103] Therefore, through the above content, the RUL prediction method provided in this embodiment has a strong correlation with the HF extracted and screened based on the IC curves of three types of batteries from NASA, and the prediction accuracy gradually increases with the increase of training data; in addition, the first 70% of the data of three types of batteries are used as training sets, and compared with ELM, BWO-ELM and LSTM respectively. The verification results show that the SABO-ELM model proposed in this embodiment has higher accuracy in predicting RUL, with a maximum root mean square error of no more than 2% and a maximum absolute error of no more than 1.5%.

[0104] Embodiment 2

[0105] This embodiment provides a lithium-ion battery remaining life prediction system based on the SABO-ELM model, including:

[0106] A health factor extraction module is used to obtain a power increment curve of the lithium-ion battery, and extract the health factor through the power increment curve;

[0107] Specifically, after collecting the capacity differential curve and the voltage differential curve during the constant current charging process of the lithium-ion battery, the charge increment curve is obtained according to the capacity differential curve and the voltage differential curve; the health factor includes: dQ / dV peak value, dQ / dV peak position and dQ / dV value other than the dQ / dV peak value.

[0108] Specifically, the Kalman filter is first performed on the power increment curve, and then the health factor in the power increment curve is extracted.

[0109] A significant health factor screening module is used to screen out significant related health factors after evaluating the correlation strength between the health factors and the capacity of the lithium-ion battery;

[0110] Specifically, the Spearman method was used to evaluate the strength of the correlation between health factors and lithium-ion battery capacity.

[0111] The extreme learning machine optimization prediction module is used to optimize the weights and bias thresholds of the extreme learning machine by using a subtraction mean algorithm, input the significantly correlated health factors into the extreme learning machine for training to obtain a SABO-ELM model, and predict the remaining life of the lithium-ion battery by using the SABO-ELM model.

[0112] Specifically, the steps of optimizing the weights and bias thresholds of the extreme learning machine using the subtraction mean algorithm include:

[0113] After setting the population size and the number of iterations and assigning the initial positions of the search agents, the objective function value of the initial positions is calculated;

[0114] The suggested position of the search agent is calculated by using a subtraction mean algorithm, and the objective function value of the suggested position is calculated. If the objective function value of the suggested position is better than the objective function value of the initial position, the initial position of the search agent is updated to the suggested position, otherwise it is not updated, and the above steps are repeated until the number of iterations is met, and the suggested position of the search agent that is finally updated is output;

[0115] The population size is the number of search agents, and the positions of the search agents are the weights and bias thresholds of the extreme learning machine.

[0116] The division of modules in the embodiments of the present invention is schematic and is only a logical function division. There may be other division methods in actual implementation. In addition, each functional module in each embodiment of the present invention may be integrated into one processor, or may exist physically separately, or two or more modules may be integrated into one module. The above-mentioned integrated modules may be implemented in the form of hardware or in the form of software functional modules.

[0117] In this embodiment, a computer device is also provided, which includes a processor and a memory, the memory is used to store a computer program (in this embodiment, the computer program includes a computing component and an iterative component, which can perform model calculation and model update), the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, which is suitable for implementing one or more instructions, and is specifically suitable for loading and executing one or more instructions in the computer storage medium to implement the corresponding method flow or corresponding function; the processor described in the embodiment of the present invention can be used for the operation of the remaining life prediction method of lithium-ion batteries based on the SABO-ELM model.

[0118] This embodiment also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both built-in storage media in a computer device and, of course, an extended storage medium supported by the computer device. The computer-readable storage medium provides a storage space that stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by a processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of the method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model in the above embodiment.

[0119] This embodiment also provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the corresponding steps of the method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model in the above embodiment are implemented.

[0120] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0121] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0122] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0123] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0124] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model, characterized in that: The following steps are involved: Obtaining a charge increment curve of the lithium-ion battery, and extracting a health factor through the charge increment curve; After evaluating the strength of the correlation between the health factor and the capacity of the lithium-ion battery, a significantly correlated health factor is screened out; After optimizing the weights and bias thresholds of the extreme learning machine by using the subtraction mean algorithm, the significantly correlated health factors are input into the extreme learning machine for training to obtain a SABO-ELM model, and the remaining life of the lithium-ion battery is predicted by the SABO-ELM model.

2. The method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model according to claim 1, characterized in that: After collecting the capacity differential curve and the voltage differential curve during the constant current charging process of the lithium ion battery, the charge increment curve is obtained according to the capacity differential curve and the voltage differential curve.

3. The method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model according to claim 1, characterized in that: The health factors include: a dQ / dV peak value, a dQ / dV peak position, and a dQ / dV value other than the dQ / dV peak value.

4. The method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model according to claim 1, characterized in that: The Spearman method was used to evaluate the strength of the correlation between the health factors and the capacity of lithium-ion batteries.

5. The method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model according to claim 1, characterized in that: The step of optimizing the weights and bias thresholds of the extreme learning machine by using the subtraction mean algorithm specifically includes: After setting the population size and the number of iterations and assigning the initial positions of the search agents, the objective function value of the initial positions is calculated; The suggested position of the search agent is calculated by using a subtraction mean algorithm, and the objective function value of the suggested position is calculated. If the objective function value of the suggested position is better than the objective function value of the initial position, the initial position of the search agent is updated to the suggested position, otherwise it is not updated, and the above steps are repeated until the number of iterations is met, and the suggested position of the search agent that is finally updated is output; The population size is the number of search agents, and the positions of the search agents are the weights and bias thresholds of the extreme learning machine.

6. The method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model according to claim 1, characterized in that: In the step of extracting the health factor through the power increment curve, the power increment curve is first subjected to Kalman filtering, and then the health factor in the power increment curve is extracted.

7. A lithium-ion battery remaining life prediction system based on the SABO-ELM model, characterized in that: include: A health factor extraction module is used to obtain a power increment curve of the lithium-ion battery, and extract the health factor through the power increment curve; A significant health factor screening module is used to screen out significant related health factors after evaluating the correlation strength between the health factors and the capacity of the lithium-ion battery; The extreme learning machine optimization prediction module is used to optimize the weights and bias thresholds of the extreme learning machine by using a subtraction mean algorithm, input the significantly correlated health factors into the extreme learning machine for training to obtain a SABO-ELM model, and predict the remaining life of the lithium-ion battery by using the SABO-ELM model.

8. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable in the processor, wherein when the processor executes the computer program, the steps of the method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model according to any one of claims 1 to 6 are implemented.

10. A computer program product, the computer product comprising a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for predicting the remaining life of a lithium-ion battery based on the SABO-ELM model according to any one of claims 1 to 6 are implemented.

Images