A constant-current charging time-based fast lithium-ion battery state of health estimation method, system and device

Abstract

The application discloses a kind of fast lithium ion battery health state estimation method, system and equipment based on constant current charging time, mainly including battery aging data set analysis, best interval screening and feature extraction, offline modeling training and evaluation, data sampling and online SOH estimation 4 steps.The features required by the method are less, the feature acquisition is easy, the model structure is simple and the parameter adjustment is easy, the model training length and SOH actual estimation length are very short, and very high SOH estimation accuracy can be achieved without any data preprocessing, which can replace the existing SOH estimation technology in real scene, and is very easy to apply.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of electric vehicle or robot battery management, and particularly relates to a fast lithium ion battery state of health estimation method, system and equipment based on constant current charging time. BACKGROUND

[0002] Battery state of health (SOH) estimation is one of the key technologies of battery management system, and the battery SOH is usually defined as the percentage between the current available capacity and the initial capacity of the battery, and real-time accurate SOH estimation is crucial for the safe operation of the battery. In recent years, data-driven methods have become a research hotspot for SOH estimation due to their flexibility and universality.

[0003] However, the current data-driven SOH estimation methods all have some common deficiencies that limit their practical application, such as the selected features for characterizing battery capacity degradation are relatively complex; most methods require tedious preprocessing steps for training data, and the model structure is complex and difficult to adjust parameters; these factors further increase the training time of the model, making it difficult to deploy online. In addition, the application of some methods requires complete charging or discharging of the battery, which results in too long actual SOH estimation time (including test data sampling time) of the battery, and it costs a lot in actual testing. In addition, the SOH estimation accuracy of these methods can be further improved. SUMMARY

[0004] In order to overcome the above-mentioned shortcomings of the prior art, the purpose of the present application is to provide a fast lithium ion battery state of health estimation method, system and equipment based on constant current charging time, which requires less features, easy to obtain features, simple model structure, short training time, high accuracy and suitable for practical application.

[0005] In order to achieve the above-mentioned purpose, the technical scheme adopted by the present application is as follows:

[0006] The application discloses a fast lithium ion battery state of health estimation method based on constant current charging time, comprising the following steps:

[0007] S1: battery aging data set analysis

[0008] The battery aging data is processed, the constant current charging data of each aging cycle is extracted, and the relationship between the constant current charging time of the voltage interval and the battery aging is analyzed, and the simplification and demonstration of the aging estimation are performed from the angles of the chargeable capacity and the chargeable energy;

[0009] S2: best interval screening and charging time feature extraction

[0010] The best sampling voltage interval is screened out, the constant current charging time of the best voltage interval is extracted as the independent variable, and the state of health SOH value of the battery is extracted as the dependent variable to complete the data preparation work.

[0011] S3: Offline modeling training and evaluation

[0012] The data set is divided into a training set and a test set, a suitable machine learning model is selected, a model is established, data is trained, the trained model is used to predict and evaluate the test set, and the trained offline model and offline evaluation results are stored in a database for quick calling when estimating SOH online.

[0013] S4: Data sampling and online SOH

[0014] The initial terminal voltage state of the battery to be tested is measured, a suitable SOH estimation strategy is selected according to the initial terminal voltage state, and corresponding constant current charging operations are performed according to the SOH estimation strategy to complete the test data sampling.

[0015] The offline model stored in the database in S3 is called to predict the sampled test data to complete the online SOH estimation.

[0016] Preferably, in S1, the battery aging behavior is described from the perspective of chargeable capacity, and the specific operation is as follows:

[0017] If the same voltage change interval is taken in each aging cycle of the battery, the chargeable capacity C charge of the battery in this interval is equal to the product of the charging current and the charging time t charge_time :

[0018] C charge = I·t charge_time ;

[0019] Wherein, I is a current constant.

[0020] Preferably, in S1, the battery aging behavior is described from the perspective of chargeable energy, and the specific operation is as follows:

[0021] In the constant voltage interval of battery charging, the chargeable energy E charge is the product of the area under the charging curve and the current, and according to the trapezoidal integral formula, E charge is expressed as:

[0022]

[0023] Wherein, t0, t n are the charging start and end times respectively, V0, V n are the charging start and end voltage values respectively, and I is a current constant.

[0024] Preferably, in S2, the implementation includes the following operations:

[0025] 2.1 Set the charging voltage starting point and interval length, thereby combining to form a charging voltage segment. For each charging voltage segment, extract the charging time of the segment corresponding to each aging cycle and the SOH value at the end of the cycle, complete data preprocessing, and prepare for screening the best voltage interval;

[0026] 2.2 Use a decision tree regression model or a correlation coefficient method to score the data set corresponding to each voltage interval, and screen out the best sampling voltage interval;

[0027] 2.3 For each aging cycle of the battery, extract the charging time corresponding to the best voltage interval and the SOH value at the end of the cycle, respectively as the independent variable and the dependent variable, thereby forming a training data set.

[0028] Preferably, in S3, the implementation includes the following operations:

[0029] 3.1 Use the leave-one-out cross-validation strategy to divide the data set into a training set and a test set;

[0030] 3.2 Use a linear regression model or a random forest model to train the training set;

[0031] 3.3 Use the trained model to predict the test set;

[0032] 3.4 Use evaluation criteria to evaluate the prediction effect of the model, and the evaluation criteria include one or more of the following indicators: mean absolute error, root mean square error, maximum absolute error, and model determination coefficient R 2

[0033] 3.5 Save the evaluation results and the trained offline model in the database for easy online estimation, and the offline evaluation results are used to guide the online estimation strategy, and the best estimation strategy is selected according to the time or accuracy requirement; the trained offline model can be directly called and online predicted in online estimation.

[0034] Preferably, in S4, the implementation includes the following operations:

[0035] 4.1 Measure the initial terminal voltage state of the battery to be tested;

[0036] 4.2 According to the initial terminal voltage of the battery, based on the offline evaluation results of S3, and combined with the estimation requirements, select a suitable SOH estimation strategy;

[0037] 4.3 According to the SOH estimation strategy, data sampling of the corresponding charging voltage segment is performed, that is, a constant current charging operation is performed and data is recorded; ​

[0038] 4.4 Call the S3 saved offline training model to predict the sampling data, and the model outputs the SOH value of the current measured battery, thereby completing the online SOH estimation.

[0039] The application further discloses a system adopting the fast lithium ion battery health state estimation method based on constant current charging time.

[0040] The battery aging data set analysis module is used for analyzing the constant current charging curve of the battery aging data set;

[0041] The optimal interval screening and charging time feature extraction module is used for screening the optimal sampling voltage interval, extracting the independent variable and the dependent variable, and completing the data preparation work.

[0042] The offline modeling training and evaluation module is used for dividing the data set, establishing a suitable machine learning model, training the data and performing SOH estimation and evaluation.

[0043] The data sampling and online SOH module is used for completing data sampling and online SOH estimation.

[0044] The application further discloses a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor realizes the steps of the fast lithium ion battery health state estimation method based on constant current charging time when executing the computer program.

[0045] Compared with the prior art, the application has the following beneficial effects:

[0046] The fast lithium ion battery health state estimation method based on constant current charging time mainly comprises four steps of battery aging data set analysis, optimal interval screening and feature extraction, offline modeling training and evaluation, and data sampling and online SOH estimation. The method requires fewer features, the features are easy to obtain, the model structure is simple and easy to tune, the model training time and the actual SOH estimation time are very short, and high SOH estimation accuracy can be achieved without any data preprocessing. The method can replace the existing SOH estimation technology in real scenes and is very easy to apply. The specific advantages are as follows:

[0047] (1) Based on the analysis of the constant current charging curve of the battery, the charging time of the fixed voltage section is used to represent the capacity degradation of the battery, and the feature is easy to obtain in practice, and the acquisition error is small.

[0048] (2) When performing online SOH estimation on the measured battery, only 0.01V of constant current charging of the battery is needed, and data sampling can be completed within 5 minutes to realize fast and accurate SOH estimation.

[0049] (3) Taking the Oxford battery aging dataset as an example, the cross-validation method is used, the average root mean square error of SOH estimation of 8 batteries is 0.56%, the maximum absolute error is less than 0.94%, and the total time of model training and prediction is less than 1 second. BRIEF DESCRIPTION OF DRAWINGS

[0050] Figure 1 A flowchart of the fast lithium-ion battery state of health estimation method based on constant current charging time of the application;

[0051] Figure 2 A constant current charging curve diagram of the Oxford soft package battery aging process in the verification case of the application;

[0052] Figure 3 An effect diagram of screening the best voltage sampling interval of the battery in the verification case of the application;

[0053] Figure 4 A mapping relationship diagram between the constant current charging time feature and the battery SOH in the verification case of the application; wherein (a) is the [3.76, 3.86]V charging time feature; (b) is the [3.81, 3.82]V charging time feature;

[0054] Figure 5 A final estimation result diagram of the battery fast online SOH estimation method based on constant current charging time of the application; wherein (a) is the SOH estimation result of the Oxford dataset 1# battery; (b) is the SOH estimation result of the Oxford dataset 2# battery. DETAILED DESCRIPTION

[0055] In order for those skilled in the art to better understand the application scheme, the technical solutions in the embodiments of the application will be described clearly and completely in conjunction with the drawings in the embodiments of the application. Obviously, the described embodiments are only a part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor should be within the scope of protection of the application.

[0056] It is to be understood that the terminology "first", "second", and the like used in the specification and the claims of the application as well as the appended drawings is merely used for distinguishing between similar objects and does not necessarily imply a required or a particular order. It is to be understood that the data used herein can be interchanged, under appropriate circumstances, so that the embodiments of the application described herein can be carried out in other than the order described herein without departing from the scope of the application. Additionally, the terms "comprising", "having", "including" and "containing" are to be construed as open-ended terms (i.e., meaning "including, but not limited to,") unless otherwise noted. It is intended that the application encompass each and every combination of features described herein.

[0057] The application will be further described in conjunction with the drawings:

[0058] As Figure 1 shown, the application provides a fast lithium-ion battery state of health estimation method based on constant current charging time, including the following steps:

[0059] Step 1), analysis of constant current charging curve of battery aging data set

[0060] According to the lithium battery cycle aging experiment or public data set, the battery aging data is obtained, and then the constant current charging data of each aging cycle is extracted, the relationship between the constant current charging time of the voltage interval and the battery aging is analyzed, and finally the simplification and demonstration of the aging estimation are carried out from the angles of the chargeable capacity and the chargeable energy.

[0061] The specific implementation is as follows:

[0062] The aging behavior of the battery is described from the angle of the chargeable capacity. If the same voltage change interval is taken in each aging cycle of the battery, the chargeable capacity of the battery in the interval is equal to the product of the charging current and the charging time:

[0063] C charge = I · t charge_time

[0064] It is noted that the battery is charged in a constant current manner, and when the charging rate is fixed, I in the above formula is a constant, and then the chargeable capacity in the constant charging voltage interval is only related to the charging time t charge_time .

[0065] The aging behavior of the battery is described from the angle of the chargeable energy. In the constant voltage interval of the battery charging, the chargeable energy E charge is the product of the area below the charging curve and the current, and according to the trapezoidal integral formula, E charge can be expressed as:

[0066]

[0067] where t0, t n are the start and end time of charging, V0, V n are the start and end voltage of charging, I is the current constant. From the above formula, it can be seen that within a fixed certain charging voltage interval, the chargeable energy of the battery is proportional to the charging time of the process, so the constant current charging time can replace the chargeable energy to represent the capacity degradation behavior of the battery. The constant current charging time feature solves the error accumulation problem that may be caused by the chargeable energy integral formula, greatly reduces the burden of the acquisition system, and is easy to obtain in practice.

[0068] Step 2), extracting the constant current charging time feature as the health state indicator

[0069] Based on the analysis of the constant current charging curve of the battery in different aging stages in step 1), the charging time in different charging voltage intervals has different corresponding relationships with the battery aging; the best voltage interval is screened out by decision tree regression or correlation coefficient method; finally, the constant current charging time of the best voltage interval is extracted as the independent variable, and the state of health (SOH) value of the battery is extracted as the dependent variable, so as to complete the data preparation and lay a foundation for subsequent modeling estimation. The specific operation steps are as follows:

[0070] Step 2.1), set the charging voltage starting point and interval length, thereby combining to form a charging voltage segment, for each voltage segment, extract the charging time of each aging cycle corresponding to the segment and the SOH value at the end of the cycle, complete data preprocessing, and prepare for screening the best voltage interval.

[0071] Step 2.2), use the decision tree regression model or the correlation coefficient method to score the data set corresponding to each voltage interval, and the voltage interval with a high score indicates that the quality of the charging time feature is higher, which is more conducive to the estimation of the battery SOH, thereby the best sampling voltage interval can be screened out.

[0072] Step 2.3), for each aging cycle of the battery, extract the charging time corresponding to the best voltage interval and the SOH value at the end of the cycle, respectively as the independent variable and the dependent variable, thereby forming a data set to be trained.

[0073] Step 3), reasonably divide the data set, establish a suitable machine learning model, train the data and perform SOH estimation and evaluation

[0074] According to the leave-one-out cross-validation strategy, the data set is divided into a training set and a test set, and then a suitable machine learning model such as a linear regression model or a random forest model is selected to establish the model, train the data and adjust the parameters (optional). Finally, the trained model is used to predict the test set, and the evaluation criteria are used to evaluate the model, and the trained model and evaluation results are stored in the database.

[0075] The specific operation steps are as follows:

[0076] Step 3.1), adopt the leave-one-out cross-validation strategy to divide the data set into a training set and a test set.

[0077] Step 3.2), adopt a linear regression model or a random forest model or other machine learning model to train the training set, and adjust the parameters appropriately according to the data fitting situation (optional).

[0078] Step 3.3), use the trained model to predict the test set.

[0079] Step 3.4), use certain evaluation criteria to evaluate the prediction effect of the model, such as mean absolute error (MAE), root mean square error (RMSE), maximum absolute error (MaxAE) and model determination coefficient R 2 and other indicators.

[0080] Step 3.5), save the SOH evaluation results in the effect evaluator and save the trained offline model in the offline model library for easy online estimation. The offline evaluation results can be used to guide the online estimation strategy, and the best estimation strategy can be selected according to the time or accuracy requirements; the trained model can be directly called and used for online prediction in online estimation.

[0081] Step 4), online SOH estimation

[0082] First, measure the initial state of the battery to be tested, and select the appropriate SOH estimation strategy according to the initial state and the effect evaluator. Then, perform the corresponding constant current charging operation according to the estimation strategy, and complete the data sampling. Finally, call the corresponding offline model saved in step 3) to predict the test data, thereby completing the online SOH estimation. The specific operation steps are as follows:

[0083] Step 4.1), measure the initial state of the battery to be tested. According to the method, the initial terminal voltage of the battery is mainly measured.

[0084] Step 4.2), according to the initial terminal voltage of the battery, based on the offline evaluation results of step 3), and combined with the estimation requirements, select the most appropriate SOH estimation strategy, such as time priority or accuracy priority strategy.

[0085] Step 4.3), data sampling of corresponding charging voltage segment is performed according to the estimation strategy, i.e. a constant current charging operation is performed and data is recorded.

[0086] Step 4.4), the offline trained model saved in step 3) is called to predict the sampled data, and the model outputs the SOH value of the currently measured battery, thus completing the online SOH estimation. Finally, the estimation result is used to guide decision-making behavior, such as determining the health level of the battery or judging whether the battery needs to be replaced.

[0087] In order to verify the feasibility and effectiveness of the method, based on the Oxford battery aging dataset, the SOH online estimation of the 1# battery in the dataset is performed, and the estimation process and results are visualized.

[0088] As shown in Figure 2 , the charging curves of the 1# battery in the Oxford dataset at different aging stages are shown. As the battery ages, the constant current charging curve moves to the left, which means that the charging time in a fixed voltage segment gradually decreases. Therefore, the constant current charging time in a specific voltage segment can to some extent represent the capacity degradation of the battery. In addition, the charging curve is extremely dense around the terminal voltage of 3.8V, which may be related to the specific internal chemical reaction of the lithium battery at this time.

[0089] As shown in Figure 3 , the SOH estimation effect of different charging voltage intervals is obtained by taking 0.01V as the starting interval of the charging voltage, and the R 2 score is represented on the vertical axis, with a higher score indicating better data fitting and higher feature quality in the interval. The 0.1V interval represented by the red dot achieves the highest R 2 score of 0.9951 when the voltage starting point is 3.76V. However, when the voltage starting point is 3.81V, the 0.01V voltage interval represented by the blue triangle achieves the highest R 2 score of 0.9897 in the same type, which means that only 0.01V of charging time can obtain high-quality features, meeting the requirement of SOH estimation speed. The 0.2V voltage interval represented by the white square achieves the highest R 2 score among all voltage interval combinations at 3.76V, although the feature fitting effect is slightly improved, but it pays a huge feature extraction time, affecting the performance of actual SOH estimation. In summary, the best charging intervals [3.76, 3.86]V and [3.81, 3.82]V are selected to meet the accuracy and speed requirements of lithium battery SOH estimation.

[0090] As shown in Figure 4The image shows a visualization of the charging time characteristics in the [3.76, 3.86] V and [3.81, 3.82] V ranges. (0.1V range...) Figure 4 The charging time characteristic corresponding to (a) is highly linearly correlated with SOH, while the 0.01V interval ( Figure 4 The relationship between the charging time characteristic corresponding to (b) and SOH approximates a quadratic convex curve, and both of these obvious mapping relationships are beneficial for completing SOH estimation. It is worth noting that the charging time required for the voltage range of [3.76, 3.86] V is approximately 600 s to 1300 s, while the voltage range of [3.81, 3.82] V can shorten this range to 50 s to 300 s! The maximum charging time is reduced by 1000 s, thus revealing the value of completing the SOH estimation of lithium batteries within 5 minutes in actual testing.

[0091] like Figure 5 The image shows the SOH estimation results for lithium batteries #1 and #2 from the Oxford dataset. The SOH estimation results based on the two voltage ranges [3.81, 3.82] V and [3.76, 3.86] V are marked as blue triangles (0.01 V range) and red dots (0.1 V range), respectively. Figure 5As can be seen from (a) and (b), both methods can well follow the degradation trend of the battery SOH, wherein the red dot representing the method based on charging 0.1V is obviously better than the estimation effect of the method based on charging 0.01V, because it sacrifices the collection or test time to strive for higher SOH estimation accuracy, but even so, it still only uses a charging voltage interval of 0.1V, which is smaller than the test interval used by most current SOH estimation methods. The method based on charging 0.01V shortens the battery test time to 300s and also maintains a high SOH estimation level, and the SOH estimation accuracy is better than the current popular method, using only 0.01V of charging data. For 1# battery, the root mean square errors of SOH estimation based on two intervals [3.81, 3.82]V and [3.76, 3.86]V are 0.67% and 0.38%, respectively, and the maximum absolute errors are 2.28% and 0.91%, respectively. For 2# battery, the root mean square errors of SOH estimation based on two intervals [3.81, 3.82]V and [3.76, 3.86]V are 1.24% and 0.78%, respectively, and the maximum absolute errors are 3.35% and 2.96%, respectively. In addition, it is worth noting that the SOH estimation method of the present application shows good estimation effect on all 8 batteries in the Oxford data set, and the average absolute errors of SOH estimation based on two intervals [3.81, 3.82]V and [3.76, 3.86]V are less than 1% on all 8 batteries. Since the leave-one-out cross-validation (the data of the tested battery does not appear in the training set) is used, it is sufficient to prove that the SOH estimation method disclosed in the present application has good robustness. In summary, the SOH estimation accuracy of the method of the present application can fully meet the actual application, and with its simple and fast characteristics, it is easy to be deployed and popularized online in practice, providing a new and efficient solution for battery health state evaluation.

[0092] In summary, the rapid lithium ion battery health state estimation method based on constant current charging time disclosed in the present application comprises four steps of battery aging data set analysis, optimal interval screening and feature extraction, offline modeling training and evaluation, data sampling and online SOH estimation. The present application can realize rapid and accurate SOH estimation through a constant current charging operation in a specific voltage interval, overcoming the problems of current SOH estimation techniques such as high feature extraction difficulty, long time consumption, complicated data preprocessing, complex model and large calculation amount. The present application can complete test data sampling and SOH online estimation in a very short time, and is very easy to apply to the actual application of SOH estimation in real scenarios.

[0093] As used herein, and unless otherwise indicated, all terms have their ordinary meanings. It should be understood that any definitions are to serve only as a tool for interpreting the description of the invention and are in no way to limit the scope thereof. As used herein, and unless otherwise indicated, the use of "or" means "and / or", the use of "and" means "and / or", and the use of "one of' means "one, two, three or more". As used herein, and unless otherwise indicated, the use of "comprise", "comprises" or "comprising" means "open ended" and does not exclude additional steps, structures or elements. As used herein, and unless otherwise indicated, the use of "or" means "and / or", the use of "and" means "and / or", and the use of "one of' means "one, two, three or more". As used herein, and unless otherwise indicated, the use of "comprise", "comprises" or "comprising" means "open ended" and does not exclude additional steps, structures or elements.

[0094] The above merely illustrates the technical idea of the present application, and cannot be used to limit the protection scope of the present application. Any modification made according to the technical idea of the present application on the basis of the technical solution falls within the protection scope of the present application.

Claims

1. A fast lithium-ion battery state-of-health estimation method based on constant-current charge time, characterized in that, Comprising the following steps: S1: battery aging data set analysis Data processing is performed on the battery aging data, the constant current charging data of each aging cycle is extracted, the relationship between the constant current charging time of the voltage interval and the battery aging is analyzed, and the simplification and demonstration of the aging estimation are performed from the angles of the chargeable capacity and the chargeable energy; Wherein, the battery aging behavior is described from the angle of the chargeable capacity, and the specific operation is as follows: If the same voltage change interval is taken in each aging cycle of the battery, the chargeable capacity of the battery in this interval is equal to the product of the charging current and the charging time . ； wherein I is a current constant; The battery aging behavior is described from the angle of the chargeable energy, and the specific operation is as follows: The energy that can be charged in the constant voltage interval of the battery charging is the product of the area under the charging curve and the current, according to the trapezoidal integration formula, is expressed as: ； wherein, , are the charging start and end times, respectively, , are the charging start and end voltage values, respectively, I is the current constant; S2: best interval screening and charging time feature extraction The best sampling voltage interval is screened out, the constant current charging time of the best voltage interval is extracted as the independent variable, and the state of health SOH value of the battery is extracted as the dependent variable, and the data preparation work is completed; The specific implementation includes the following operations: 2.1 Set the charging voltage starting point and interval length, thereby combining to form a charging voltage segment, for each charging voltage segment, extract the charging time of the segment corresponding to each aging cycle and the SOH value at the end of the cycle, complete the data preprocessing, and prepare for screening the best voltage interval; 2.2 Use a decision tree regression model or a correlation coefficient method to score the data set corresponding to each voltage interval, and screen out the best sampling voltage interval; 2.3 For each aging cycle of the battery, extract the charging time corresponding to the best voltage interval and the SOH value at the end of the cycle as the independent variable and the dependent variable respectively, thereby forming a training data set; S3: offline modeling training and evaluation Divide the data set into a training set and a test set, select a suitable machine learning model, establish a model, train the data, use the trained model to predict and evaluate the test set, store the trained offline model and offline evaluation results in the database for quick calling when estimating SOH online; S4: data sampling and online SOH Measure the initial terminal voltage state of the battery to be tested, select a suitable SOH estimation strategy according to the initial terminal voltage state, perform the corresponding constant current charging operation according to the SOH estimation strategy, and complete the test data sampling; Call the offline model stored in the database in S3 to predict the sampled test data and complete the online SOH estimation; The specific implementation includes the following operations: 4.1 Measure the initial terminal voltage state of the battery to be tested; 4.2 Select a suitable SOH estimation strategy based on the offline evaluation results of S3 according to the initial terminal voltage of the battery and in combination with the estimation requirements; 4.3 Perform data sampling of the corresponding charging voltage segment according to the SOH estimation strategy, that is, perform a constant current charging operation and record the data; 4.4 Call the offline training model saved in S3 to predict the sampled data, and the model outputs the SOH value of the battery being tested, thereby completing the online SOH estimation.

2. The constant current charge time based fast lithium-ion battery state of health estimation method of claim 1, wherein, In S3, the specific implementation includes the following operations: 3.1 Use the leave-one-out cross-validation strategy to divide the data set into a training set and a test set; 3.2 Use a linear regression model or a random forest model to train the training set; 3.3 Use the trained model to predict the test set; 3.4 The prediction performance of the model is evaluated using evaluation criteria including mean absolute error, root mean square error, maximum absolute error, and model determination coefficient one or more of the indicators; 3.5 The evaluation results and the trained offline model are saved in the database, which is convenient for fast calling in online estimation. The offline evaluation results are used to guide the online estimation strategy, and the best estimation strategy is selected according to the time or accuracy requirement; The trained offline model can be directly called and online prediction is performed in online estimation.

3. A system employing the fast lithium-ion battery state-of-health estimation method based on constant current charge time according to any one of claims 1-2, characterized in that, It comprises: a battery aging data set analysis module for analyzing the constant current charging curve of the battery aging data set; a best interval screening and charging time feature extraction module for screening the best sampling voltage interval, extracting the independent variable and the dependent variable, and completing the data preparation work; an offline modeling training and evaluation module for dividing the data set, establishing a suitable machine learning model, training the data, and performing SOH estimation and evaluation; a data sampling and online SOH module for completing data sampling and online SOH estimation.

4. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, The processor implements the steps of the fast lithium ion battery health state estimation method based on constant current charging time according to any one of claims 1-2 when executing the computer program.

Images