A Lithium Battery Life Prediction Method Based on the Mean Range of Degradation Quantities

Abstract

The present invention discloses a lithium battery life prediction method based on the mean range of degradation amount. By accelerating the life experiment of the lithium battery, the capacities of the sample lithium battery and the lithium battery to be tested at different times are obtained respectively. Then, the sample lithium batteries that meet the floating range of the mean value are screened out by means of the mean range of degradation amount. Next, the mean prediction error is calculated by using the ELM network, and the predicted value of the capacity degradation amount is corrected by correcting the mean error until the corrected capacity degradation amount reaches the failure threshold, so as to obtain the remaining life of the lithium battery to be tested.

Description

Technical Field

[0001] The present invention belongs to the technical field of battery reliability analysis. More specifically, it relates to a method for predicting the life of a lithium battery based on the mean range of degradation amounts. Background Art

[0002] Lithium batteries are an important energy storage device, widely used in many key fields such as electronic products, new energy vehicles, aerospace, etc. Their advantages include high efficiency, portability, and fast charging. However, limited by factors such as manufacturing processes, usage methods, and external environments, the effective capacity of lithium batteries, that is, the total electrical energy that can be provided from a full charge to a complete discharge, will continuously decrease with the usage time, and the decrease value is called the capacity degradation amount. Such degradation will have a certain impact on the stability of system operation, mainly reflected in the shortening of the operation cycle and the power shortage caused by voltage drop. Therefore, the research on the prediction technology of the remaining useful life of the battery module has become very necessary. This research aims to find the optimal prediction method to make the prediction result closest to the real battery life, so as to achieve the following goals: (1) As an important way to obtain the reliability information of the lithium battery module, more accurately obtain the information of battery data and health status; (2) Better design the accelerated aging test to obtain more accurate aging data, so as to give early reminders and warnings; (3) Help reduce the investment in lithium battery stability detection, conduct unified maintenance at the appropriate time according to the prediction data, and reduce production costs; (4) As a prediction idea, it can be extended to other similar fields to improve the prediction efficiency and effect in other fields.

[0003] The remaining useful life of a lithium battery refers to the number of charge-discharge cycles required for the maximum available capacity of the battery to decay to a certain specified failure threshold after a certain charge-discharge process. The existing methods for predicting the life of lithium batteries mainly include two categories: based on physical failure models and data-driven. Model-based prediction methods usually require in-depth understanding of the internal structure, material properties, aging mechanism, etc. of the battery, are greatly limited by external environmental factors, and often only target a specific battery system, with a certain accuracy. However, the established model is not universal, and due to the complex chemical reactions inside the battery, the establishment process is also more cumbersome. Data-driven prediction methods do not require in-depth study of the aging mechanism and rules of the battery itself, but establish statistical models or machine learning models based on data, which are more suitable for life prediction in different occasions and situations. For data-driven prediction methods, the powerful computing power of the machine does improve the accuracy to a certain extent. However, the prediction accuracy depends on a large data basis and good data quality. Summary of the Invention

[0004] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a lithium battery life prediction method based on the mean range of degradation amounts, which samples the mean range of degradation amounts to correct the error of the predicted value of the capacity degradation amount of the lithium battery to be measured at subsequent moments, improves the prediction accuracy, and thus ensures the reliability of the lithium battery operation.

[0005] To achieve the above invention purpose, a lithium battery life prediction method based on the mean range of degradation amounts of the present invention is characterized by including the following steps:

[0006] (1) Obtain the capacities of the sample lithium batteries at different moments;

[0007] By accelerating the life experiments of N sample lithium batteries, sampling the capacities of each sample lithium battery at different moments, and then subtracting the initial capacity from the capacity of each sample lithium battery at different moments, the capacity degradation amount of each sample lithium battery at different moments is obtained. Among them, the capacity degradation amount of the i-th sample lithium battery at different moments is denoted as where i = 1, 2,..., N, represents the initial capacity degradation amount of the i-th sample lithium battery, represents the capacity degradation amount of the i-th sample lithium battery at the t-th moment, and T represents the number of data sampling moments;

[0008] (2) Obtain the capacities of the lithium battery to be measured at the first t moments, where t << T;

[0009] By accelerating the life experiment of the lithium battery to be measured, sampling the capacities of the lithium battery to be measured at the first t moments, and then subtracting the initial capacity from the capacities of the lithium battery to be measured at different moments, the capacity degradation amount L = {L0, L1,..., L t} of the lithium battery to be measured at different moments is obtained. Among them, L0 represents the initial capacity degradation amount of the lithium battery to be measured, and L t represents the capacity degradation amount of the lithium battery to be measured at the t-th moment;

[0010] (3) Construct a sample set;

[0011] (3.1) Traverse the capacity degradation amount of each sample lithium battery. Among them, in the capacity degradation amount of the i-th sample lithium battery, take the last t / 4 items of data to calculate their mean value, denoted as

[0012] When the capacity degradation amounts of N sample lithium batteries are traversed, a mean value sequence {μ1, μ2,..., μ i ,..., μ N} is obtained;

[0013] (3.2) Select the maximum mean value and the minimum mean value in the mean value sequence, and then calculate their difference to obtain the mean range of degradation amounts, denoted as Rμ ;

[0014] (3.3) The capacity degradation of the lithium battery under test at the first t moments is L = {L0, L1, ..., L t}, take the last t / 4 items of data and calculate their mean, which is recorded as μ=4(L 3t / 4+1 +L 3t / 4+2 +L 3t / 4+3 +…+L t ) / t;

[0015] (3.4) Set the upper and lower floating range of the mean R μ / n, n is a floating parameter;

[0016] Traverse the mean sequence {μ1,μ2,…,μ i ,…,μ N}, select the mean that satisfies the following formula;

[0017] μ-R μ / n≤μ i ≤μ+R μ / n

[0018] (3.5) Assume that a total of K eligible means are screened out, denoted as {μ1,μ2,…,μ j ,…,μ K}; The capacity degradation of the sample lithium batteries corresponding to these K means is combined into a sample set, which is recorded as {L1, L2, ..., L j ,…,L K};

[0019] (4) Calculate the mean prediction error using the ELM network;

[0020] (4.1), in the mean sequence {μ1,μ2,…,μ j ,…,μ K}, taking μ1 as the reference, in μ2~μ K Select the mean closest to μ1 and record it as μ k , the corresponding capacity degradation is recorded as

[0021] (4.2), in L k Extract the capacity degradation amount of the previous t moments Then take the last p data and input them into the trained ELM network to obtain the predicted value of capacity degradation at time t+1

[0022] (4.3) Calculate the prediction error of the kth sample lithium battery at time t+1

[0023]

[0024] (4.4) Similarly, by traversing each mean value in {μ1, μ2, …, μ j , …, μ K}, the prediction errors of the K sample lithium batteries at the (t + 1)-th moment are obtained according to the methods described in steps (4.1)-(4.3), and an error sequence at the (t + 1)-th moment is formed

[0025] (4.5) Calculate the mean value of the prediction errors of the K sample lithium batteries at the (t + 1)-th moment, Δμ t+1 ;

[0026]

[0027] (4.6) Calculate the mean values of the prediction errors of the K sample lithium batteries at the (t + 2)-th, (t + 3)-th to T-th moments in turn according to the methods described in steps (4.2)-(4.5), and obtain a sequence of mean values of prediction errors {Δμ t+1 , Δμ t+2 , …, Δμ T};

[0028] (5) Use the difference to correct the prediction result;

[0029] (5.1) Among the capacity degradations L = {L0, L1, …, L t} of the lithium battery to be measured at the first t moments, take the last p data and input them into the trained ELM network, so as to obtain the predicted value of the capacity degradation at the (t + 1)-th moment

[0030] (5.2) Calculate the mean value of the capacity degradations of the K sample lithium batteries at the (t + 1)-th moment, ΔL t+1 ;

[0031]

[0032] (5.3) Correct the mean value of the prediction error Δμ according to the ratio of the predicted value of the capacity degradation of the lithium battery to be measured at the (t + 1)-th moment t+1 to the mean value ΔL t+1 ;

[0033]

[0034]

[0035] Wherein, represents the corrected mean value of the prediction error at the (t + 1)-th moment;

[0036] (5.4), Modify the predicted value of the capacity degradation of the lithium battery under test at time t + 1

[0037]

[0038] Among them, represents the corrected capacity degradation at time t + 1;

[0039] (5.5), Judge whether it is less than the given failure threshold w. If it is less, go to step (6) and the algorithm ends; otherwise, go to step (5.6);

[0040] (5.6), Incorporate the corrected capacity degradation at time t + 1 into the capacity degradation sequence;

[0041]

[0042] (5.7), According to the method described in steps (5.1)-(5.6), sequentially correct the predicted values of the capacity degradation at times t + 2, t + 3 to T;

[0043] (6), The capacity of the lithium battery under test degrades to the failure threshold w, the prediction terminates, and the algorithm ends.

[0044] The invention purpose of the present invention is realized as follows:

[0045] Based on the lithium battery life prediction method of degradation amount mean range, by accelerating the life experiment of the lithium battery, the capacities of the sample lithium battery and the lithium battery under test at different times are respectively obtained, and then the sample lithium batteries that meet the floating range of the mean value are screened out by means of the degradation amount mean range; then the mean value of the prediction error is calculated by using the ELM network, and the predicted value of the capacity degradation is corrected by correcting the mean value of the error until the corrected capacity degradation reaches the failure threshold, so as to obtain the remaining life of the lithium battery under test.

[0046] At the same time, the lithium battery life prediction method based on the degradation amount mean range of the present invention also has the following beneficial effects:

[0047] (1), By using the sample data screening method based on the mean range, the present invention helps to reduce the influence of the sample group with large error deviation on the prediction result and improves the prediction accuracy of the model;

[0048] (2), The present invention adopts the method of statistical correction of single-step error to reduce the influence of cumulative error on the experimental result;

[0049] (3) The present invention uses a correction factor based on the mean difference value to correct the prediction result. Compared with the traditional ELM, the mean square error of the remaining life predicted by the present invention is significantly smaller than that of the traditional ELM. In this way, the present invention can better reduce the prediction error and improve the prediction accuracy of the remaining life. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is a flowchart of the lithium battery life prediction method based on the mean range of degradation amount of the present invention;

[0051] Figure 2 is a prediction curve of the remaining life of the lithium battery to be tested predicted by the present invention;

[0052] Figure 3 is a prediction curve of the remaining life of the lithium battery to be tested predicted based on the Extreme Learning Machine (ELM). DETAILED DESCRIPTION OF THE INVENTION

[0053] The following describes the specific implementation manners of the present invention with reference to the drawings, so that those skilled in the art can better understand the present invention. It should be particularly noted that in the following description, when the detailed description of known functions and designs may dilute the main content of the present invention, these descriptions will be omitted here.

[0054] Embodiment

[0055] Figure 1 is a flowchart of the lithium battery life prediction method based on the mean range of degradation amount of the present invention.

[0056] In this embodiment, as Figure 1 shown, a lithium battery life prediction method based on the mean range of degradation amount of the present invention includes the following steps:

[0057] S1. Obtain the capacities of the sample lithium batteries at different times;

[0058] By accelerating the life experiments of N sample lithium batteries, sampling the capacities of each sample lithium battery at different times, and then subtracting the capacity of each sample lithium battery at different times from the initial capacity, the capacity degradation amount of each sample lithium battery at different times is obtained. Among them, the capacity degradation amount of the i-th sample lithium battery at different times is denoted as where i = 1, 2,..., N, represents the initial capacity degradation amount of the i-th sample lithium battery, represents the capacity degradation amount of the i-th sample lithium battery at the t-th time, and T represents the number of data sampling times;

[0059] S2. Obtain the capacities of the lithium battery to be tested at the first t times, where t << T;

[0060] By accelerating the life experiment of the lithium battery to be tested, sampling the capacity of the lithium battery to be tested at the first t moments, and then subtracting the capacity of the lithium battery to be tested at different moments from the initial capacity, the capacity degradation amount L of the lithium battery to be tested at different moments is obtained = {L0, L1, …, L t}, where L0 represents the initial capacity degradation amount of the lithium battery to be tested, and L t represents the capacity degradation amount of the lithium battery to be tested at the t-th moment;

[0061] In this embodiment, 14 groups of sample lithium batteries are selected, with models CS2-24, CS2-25, CS2-26, CS2-27, CS2-28, CS2-29, CS2-30, CS2-31, CS2-32, CS2-33, CS2-34, CS2-35, CS2-36, CS2-38. The lithium battery CS2-37 to be tested is selected as the prediction object, and the capacity degradation amounts of the sample lithium battery and the lithium battery to be tested at different moments are obtained respectively according to the above method;

[0062] S3. Construct a sample set;

[0063] In the actual prediction process, we often find that due to the inclusion of samples with too large numerical deviations, the prediction result error is relatively large. Therefore, a method for removing samples with too large deviations is provided here, that is, calculating the mean value of the capacity degradation amounts of each group of lithium batteries within a certain time period and selecting those close to the group to be predicted. The specific steps are as follows:

[0064] S3.1. Traverse the capacity degradation amount of each sample lithium battery. Among them, in the capacity degradation amount of the i-th sample lithium battery, take the last t / 4 items of data to calculate their mean value, denoted as

[0065] After traversing the capacity degradation amounts of N sample lithium batteries, a mean value sequence {μ1, μ2, …, μ i , …, μ N} is obtained;

[0066] S3.2. Select the maximum mean value and the minimum mean value in the mean value sequence, and then calculate their difference to obtain the range of the mean value of the degradation amount, denoted as R μ ;

[0067] In this embodiment, the capacity degradation amount data in the time period from moment 301 to 400 is selected to calculate the mean value, where the maximum is 0.16637, the minimum is 0.1409, and the range is 0.0255.

[0068] S3.3. In the capacity degradation amount L = {L0, L1, …, L t} of the lithium battery to be tested at the first t moments, take the last t / 4 items of data to calculate their mean value, denoted as μ = 4(L3t / 4+1 +L 3t / 4+2 +L 3t / 4+3 +…+L t ) / t;

[0069] S3.4. Set the mean upper and lower floating range R μ / n, where n is a floating parameter, and its magnitude is inversely proportional to the screening range and the actual amount of data obtained by screening. The value is determined by the actual conditions and the characteristics of the data distribution. The value of n should be reasonably adjusted in different experimental scenarios to meet the requirements for the quality and capacity of the samples during the prediction process.

[0070] Traverse the mean sequence {μ1, μ2, …, μ i , …, μ N}, and select the means that satisfy the following formula;

[0071] μ - R μ / n ≤ μ i ≤ μ + R μ / n

[0072] S3.5. Assume that a total of K means that meet the conditions are selected, denoted as {μ1, μ2, …, μ j , …, μ K}; form a sample set from the capacity degradation amounts of the sample lithium batteries corresponding to these K means, denoted as {L1, L2, …, L j , …, L K};

[0073] In this embodiment, after comparing the prediction result errors for different values of n multiple times, the optimal value of n is finally determined to be 8. Therefore, under the condition of n = 8, the mean range of the effective degradation data group is from 0.1511 to 0.1575. Among them, a total of 9 groups, namely CS2-25, CS2-26, CS2-28, CS2-29, CS2-31, CS2-33, CS2-34, CS2-35, and CS2-36, meet the requirements.

[0074] S4. Calculate the mean value of the prediction error using the ELM network;

[0075] S4.1. In the mean sequence {μ1, μ2, …, μ j , …, μ K}, taking μ1 as the benchmark, select the mean in μ2 ∼ μ K that is closest in magnitude to μ1, denoted as μ k , and the corresponding capacity degradation amount is denoted as

[0076] S4.2. Extract the capacity degradation amounts at the first t time instants from L k Then, the last p data are taken and input into the trained ELM network to obtain the predicted value of the capacity degradation at the (t + 1)-th moment.

[0077] In this embodiment, the ELM network is an existing network. The structure and training process of the ELM network will not be elaborated here, and a trained ELM network is specifically selected.

[0078] S4.3. Calculate the prediction error of the k-th sample lithium battery at the (t + 1)-th moment.

[0079]

[0080] S4.4. Similarly, by traversing each mean value in {μ1, μ2, …, μ j , …, μ K}, the prediction errors of the K sample lithium batteries at the (t + 1)-th moment are obtained according to the methods described in steps S4.1 - S4.3, forming an error sequence at the (t + 1)-th moment.

[0081] S4.5. Calculate the mean value of the prediction errors of the K sample lithium batteries at the (t + 1)-th moment, Δμ t+1 ;

[0082]

[0083] S4.6. Calculate the mean values of the prediction errors of the K sample lithium batteries at the (t + 2)-th, (t + 3)-th to T-th moments in turn according to the methods described in steps S4.2 - S4.5, obtaining a sequence of mean values of prediction errors {Δμ t+1 , Δμ t+2 , …, Δμ T};

[0084] S5. Correct the prediction result using the difference;

[0085] S5.1. Among the capacity degradations L = {L0, L1, …, L t} of the lithium battery to be measured at the first t moments, take the last p data and input them into the trained ELM network to obtain the predicted value of the capacity degradation at the (t + 1)-th moment.

[0086] S5.2. Calculate the mean value of the capacity degradations of the K sample lithium batteries at the (t + 1)-th moment, ΔL t+1 ;

[0087]

[0088] S5.3. According to the predicted value of the capacity degradation of the lithium battery to be measured at the (t + 1)-th moment and the mean value ΔLt+1 Ratio-corrected predicted error mean Δμ t+1 ;

[0089]

[0090]

[0091] wherein, represents the corrected predicted error mean at time t + 1;

[0092] S5.4. Correct the predicted value of the capacity degradation of the lithium battery to be measured at time t + 1

[0093]

[0094] wherein, represents the corrected capacity degradation at time t + 1;

[0095] S5.5. Judge whether it is less than the given failure threshold w. If it is less, go to step (6) and the algorithm ends; otherwise, go to step S5.6;

[0096] S5.6. Incorporate the corrected capacity degradation at time t + 1 into the capacity degradation sequence;

[0097]

[0098] S5.7. Correct the predicted values of the capacity degradation at times t + 2, t + 3 to T in sequence according to the method described in steps S5.1 - S5.6;

[0099] S6. When the capacity of the lithium battery to be measured degrades to the failure threshold w, the prediction terminates and the algorithm ends.

[0100] In this embodiment, 400 is selected as the reference time, and predictions are made every 10 moments. The results obtained are as Figure 2 shown. Among them, the downward-sloping straight line is the true remaining life value of the lithium battery to be measured, and the other curve is the life prediction value obtained based on the present invention. By comparing the differences at the same moment, it can be seen that the mean square error of the remaining life prediction obtained by the present invention is less than 45.

[0101] To quantitatively compare and measure the prediction performance, Figure 2It shows the prediction results of the remaining life of lithium battery CS2-37 by a general ELM network without data screening based on the mean range of degradation amount. Among them, the downward-sloping straight line is the true remaining life value of the lithium battery to be measured, and the other curve is the predicted value of the remaining life of the lithium battery based on the Extreme Learning Machine (ELM). By comparing the differences at the same moment, it can be seen that the mean square error of the prediction method of the remaining life of the lithium battery based on the Extreme Learning Machine (ELM) is as high as 76.

[0102] Table 1 shows the comparison of the mean square error of the prediction error between the present invention and ELM;

[0103] The present invention ELM Mean square error 45 76

[0104] Table 1

[0105] From the prediction results shown in Table 1, it can be seen that the accuracy of the prediction results of the remaining life of the lithium battery to be measured by the present invention is much higher than that of the general ELM model. Therefore, compared with the existing neural network prediction models, the present invention has higher prediction accuracy and is thus more suitable for the needs of predicting the remaining life of lithium batteries in practical engineering.

[0106] Although the above description of the illustrative specific embodiments of the present invention is provided for the understanding of those skilled in the art of the present technology, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art of the present technology, as long as various changes are within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions and creations using the concept of the present invention are within the scope of protection.

Claims

1. A lithium battery life prediction method based on the mean range of degradation amounts, characterized in that The following steps are involved: (1) Obtain the capacity of the sample lithium battery at different times; By accelerating the life experiments of N sample lithium batteries, sampling the capacities of each sample lithium battery at different times, and then subtracting the capacity of each sample lithium battery at different times from the initial capacity, the capacity degradation amount of each sample lithium battery at different times is obtained. Among them, the capacity degradation amount of the i-th sample lithium battery at different times is denoted as where i = 1, 2, …, N, represents the initial capacity degradation amount of the i-th sample lithium battery, represents the capacity degradation amount of the i-th sample lithium battery at the t-th time, and T represents the number of data sampling times; (2) Obtain the capacity of the lithium battery to be tested at the previous t moments, t<<T; By accelerating the life experiment of the lithium battery under test, sampling the capacity of the lithium battery under test at the first t moments, and then taking the difference between the capacity of the lithium battery under test at different moments and the initial capacity, the capacity degradation amount L = {L0, L1, …, L t} is obtained, where L0 represents the initial capacity degradation amount of the lithium battery under test, and L t represents the capacity degradation amount of the lithium battery under test at the t-th moment; (3) Constructing a sample set; (3.1) Traverse the capacity degradation of each sample lithium battery. Among them, for the capacity degradation of the i-th sample lithium battery take the mean of the last t / 4 data items, denoted as After the capacity degradation of N sample lithium batteries is traversed, the mean sequence {μ1, μ2, …, μ i , …, μ N}; (3.2) Select the maximum mean and the minimum mean in the mean value sequence, and then calculate their difference to obtain the mean range of the degradation amount, denoted as R μ ; (3.3)、In the capacity degradation amounts \(L = \{L_0, L_1, \ldots, L_{}\) of the lithium battery under test at the first \(t\) moments, take the last \(t / 4\) data items and calculate their mean value, denoted as \(\mu = 4(L_{}\) t + L_{} 3t / 4+1 + L_{} 3t / 4+2 + L_{} 3t / 4+3 + \ldots + L_{} t ) / t; (3.4) Set the mean floating range R μ / n, where n is a floating parameter; Traverse the mean sequence {μ1, μ2, …, μ i , …, μ N}, and select the mean that satisfies the following formula; μ-R μ / n ≤ μ i ≤ μ + R μ / n (3.5) Suppose a total of K eligible means are screened out, denoted as {μ1, μ2, …, μ j , …, μ K}; The capacity degradation amounts of the sample lithium batteries corresponding to these K means form a sample set, denoted as {L1, L2, …, L j , …, L K}; (4) Calculate the mean prediction error using the ELM network; (4.1) In the mean value sequence {μ1, μ2, …, μ j , …, μ K}, taking μ1 as the benchmark, select the mean value in μ2 ∼ μ K that is closest in magnitude to μ1, denoted as μ k , and the corresponding capacity degradation amount is denoted as (4.2), extract the capacity degradation amounts at the first t moments in L k Then take the last p data and input them into the trained ELM network to obtain the predicted value of the capacity degradation amount at the (t + 1)-th moment ​ (4.3) Calculate the prediction error of the k-th sample lithium battery at the (t + 1)-th moment (4.4) Similarly, by traversing each mean value in {μ1, μ2, …, μ j , …, μ K}, the prediction errors of the K sample lithium batteries at the (t + 1)-th moment are obtained according to the methods described in steps (4.1)-(4.3), and an error sequence at the (t + 1)-th moment is formed (4.5), calculate the mean prediction error Δμ of the K-sample lithium battery at the (t + 1)-th moment t+1 ; (4.6) Calculate the mean value of the prediction errors of the K sample lithium batteries at the moments of t+2, t+3 to T in sequence according to the method described in steps (4.2)-(4.5), and obtain the prediction error mean value sequence {Δμ t+1 , Δμ t+2 , …, Δμ T}; (5) Use the difference to correct the prediction results; (5.1) In the capacity degradation amounts L = {L0, L1, …, L t} of the lithium battery under test at the first t moments, take the last p data and input them into the trained ELM network, so as to obtain the predicted value of the capacity degradation amount at the (t + 1)-th moment (5.2) Calculate the mean value ΔL of the capacity degradation of K sample lithium batteries at the (t + 1)-th moment t+1 ; (5.3), according to the predicted value of the capacity degradation of the lithium battery to be measured at time t + 1 and the mean value ΔL t+1 to correct the mean prediction error Δμ t+1 ; Among them, represents the mean of the corrected prediction error at time t + 1; (5.4) Modify the predicted value of the capacity degradation of the lithium battery under test at time t + 1 Among them, represents the corrected capacity degradation at time t + 1; (5.5), Determine whether it is less than the given failure threshold w. If it is less, go to step (6) and the algorithm ends; otherwise, go to step (5.6); (5.6) Incorporate the corrected capacity degradation at time t + 1 into the capacity degradation sequence; (5.7) According to the method described in steps (5.1)-(5.6), the predicted values of capacity degradation at time t+2, t+3 to time T are corrected in sequence; (6) When the capacity of the lithium battery to be tested degrades to the failure threshold w, the prediction is terminated and the algorithm ends.

Images