Lithium ion battery health state estimation method, system, equipment and medium
By transforming the capacity increment curve into image data using a segmented aggregation approximation algorithm and a GAF algorithm, and combining 2DCNN and LSTM models, the problem of complexity and inadequacy in feature extraction in lithium-ion battery health status estimation is solved, achieving more accurate health status estimation.
Patent Information
- Application Number
- CN202411086782.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-08
- Publication Date
- 2024-11-19
AI Technical Summary
In existing technologies, the feature extraction process for estimating the state of health of lithium-ion batteries is complex and insufficient, making it difficult to achieve accurate battery state of health estimation.
The segmented aggregation approximation algorithm and Gram transform algorithm (GAF) are used to transform the capacity increment curve into image data. Combined with 2DCNN and LSTM models, features are extracted and fused from multiple dimensions to generate fused comprehensive features and train them to output health status estimation results.
It achieves more accurate estimation of the state of health of lithium-ion batteries, reduces the complexity of feature extraction and the impact of noise, and improves the accuracy and stability of estimation.
Smart Images

Figure CN118980957A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of battery management, and in particular to a method, system, device and medium for estimating the health status of a lithium-ion battery. Background Art
[0002] Lithium-ion batteries play an increasingly important role in various fields such as electronics, electric vehicles, microgrids, and energy storage systems due to their advantages such as long cycle life, high energy density, and strong reliability. However, the development of lithium-ion batteries still faces major challenges due to safety issues during operation. As a complex electrochemical system, the electrochemical properties of lithium-ion batteries will change and degrade during continuous operation. The continuous deterioration of battery performance will seriously affect its service life, resulting in a reduction in available capacity and even safety hazards. Therefore, in order to improve the performance of lithium batteries, the state of health (SOH) must be accurately estimated. At present, the SOH of the battery has not been formally defined, and any variable that can reflect the battery usage status can be used to define the SOH. In general, the SOH of a battery is usually defined as a decrease in battery capacity or an increase in internal resistance. This application regards the ratio of the current maximum available capacity to the nominal capacity as the SOH of the battery. The research on SOH estimation can be mainly divided into two categories: model-based methods and data-driven methods.
[0003] The model-based approach combines the battery model with technologies such as Kalman filters and particle filters to estimate certain parameters in the model that change as the battery ages to characterize the battery's SOH. Models that describe battery characteristics include electrochemical models and equivalent circuit models. The equivalent circuit model has limited accuracy and poor adaptability and cannot describe the internal reaction mechanism of the battery. The electrochemical model has problems such as overly complex models and difficult parameter identification. The model-based approach relies on the accuracy of the battery model. However, the aging mechanism of lithium-ion batteries is very complex, and it is always challenging to establish an accurate estimation model.
[0004] In contrast, data-driven methods do not rely on accurate battery models. Instead, they use specific learning algorithms to extract aging information from historical data points and establish a relationship between data and health status. Although data-driven methods have shown strong capabilities in the field of lithium-ion battery SOH estimation, they still have two problems: (1) Feature information is usually obtained through battery aging experiments, and the noise in the measurement process will affect feature extraction. How to choose a suitable algorithm to filter out noise from the original curve remains a challenge. At the same time, feature extraction is a complex task, and the quality of extracted features has a great impact on the accuracy of SOH estimation. (2) Since the aging of lithium-ion batteries is a very complex process, it is difficult to fully reflect the health status of lithium-ion batteries by only extracting some features during the charging and discharging process. It is necessary to extract and analyze various features from different perspectives. In the current study, battery health-related features are obtained from one-dimensional time series data, and the extracted features may not fully reflect the aging process of lithium-ion batteries.
[0005] In summary, the existing related technologies have technical problems such as complex feature extraction process and insufficient feature extraction when estimating the battery health state, making it difficult to achieve accurate battery health state estimation. Summary of the invention
[0006] In order to solve the problems existing in the prior art, the present invention provides a lithium-ion battery health status estimation method, system, device and medium, which are used to solve the technical problems in the existing related technology when estimating the battery health status, such as complex feature extraction process and insufficient feature extraction, making it difficult to achieve accurate battery health status estimation.
[0007] To achieve the above object, the present invention provides the following technical solutions:
[0008] A method for estimating the health status of a lithium-ion battery comprises the following steps:
[0009] Collect the voltage, current, capacity and corresponding sampling time data of lithium-ion batteries during the constant current charging stage of the charge and discharge cycle;
[0010] Obtaining a capacity increment curve based on the voltage, current, capacity and corresponding sampling time data;
[0011] Processing the capacity increment curve based on a segmented aggregation approximation algorithm to generate a short sequence capacity increment data curve;
[0012] The short sequence capacity increment data curve is transformed based on the GAF algorithm to generate image data;
[0013] A 2DCNN network is used to extract first dimensional features from the image data, a 1DCNN network is used to extract second dimensional features from the capacity increment curve, and the first dimensional features and the second dimensional features are fused based on a feature fusion network to generate fused comprehensive features;
[0014] The fused comprehensive features are trained based on the LSTM model to output the health status estimation result.
[0015] Furthermore, the process of obtaining the battery health status, voltage, current, capacity, and sampling time data of each charge and discharge cycle of the lithium-ion battery in the constant current charging stage is as follows:
[0016] The constant current charging voltage data of the i-th cycle is the sequence U i,0 ,U i,1 ,...,U i,j , where U i,0 and U i,j They are the initial voltage and the final voltage of the constant current charging stage respectively;
[0017] The constant current charging current data of the i-th cycle is sequence I i,0 ,I i,1 ,...,I i,j , where I i,0 and I i,j They are the initial current and the termination current in the constant current charging stage respectively;
[0018] The constant current charging sampling time data of the i-th cycle is the sequence t i,0 ,t i,1 ,...,t i,j , where t i,0 and t i,j They are the initial sampling time and the terminal sampling time of the constant current charging stage respectively;
[0019] The charging capacity data in the constant current charging stage is the sequence Q1, Q2, ..., Q n , Q i =∫I i dt i , where Q i is the constant current charging capacity of the battery in the i-th cycle, I i and t i are the current and sampling time of the i-th cycle respectively.
[0020] Furthermore, the capacity increment curve is:
[0021]
[0022] In the formula, △Q and △U represent the change in capacity and voltage of the battery from time m to time m+1, respectively, and Um+1 represents the battery voltage at time m+1, U m represents the voltage at time m, Q m represents the battery capacity at time m, Q m+1 represents the capacity of the battery at time m+1, I represents the current of the battery, and t m+1 represents the sampling time at time m+1, t m Represents the sampling time at time m.
[0023] Furthermore, the process of processing the capacity increment curve based on the segmented aggregation approximation algorithm to generate a short sequence capacity increment data curve is as follows:
[0024] Given an IC sequence Z1,Z2,...,Z of length m m , turning it into a data sequence of length w where w <m;
[0025] The resulting compression ratio is The dimensionality reduction process of time series data, and satisfy
[0026] In the formula, represents the ith IC data after dimension reduction, Z j represents the jth original IC data, is the compression ratio.
[0027] Furthermore, the process of converting the short sequence capacity increment data curve based on the GAF algorithm to generate image data is as follows:
[0028]
[0029] Among them, x i and represent the original IC sequence and the normalized IC sequence at the i-th time point, max(X) and min(X) represent the maximum and minimum values of the IC sequence, respectively; φ i represents the polar angle at the i-th time point, r i Indicates the extreme meridian, t i represents the timestamp, N is the total number of sampling points; GASF represents a Gram-like matrix that maintains time dependency.
[0030] Furthermore, the process of generating the fusion comprehensive features is as follows:
[0031]
[0032] a 2D =g(y 2D ),a 2D ∈Rm
[0033]
[0034] a 1D =g(y 1D ),a 1D ∈R n
[0035] x f =concat(a 2D ,a 1D ),x f ∈R m+n
[0036] Where W 2D and W 1D Represent the weight matrices of 2DCNN and 1DCNN respectively, a 2D and a 1D Respectively represent the feature vectors after 2DCNN and 1DCNN; g(y 1D ), g(y 2D ) represents the activation function, concat(a 2D ,a 1D ) represents the feature fusion function, x f Represents the fused feature vector.
[0037] Furthermore, the LSTM model is:
[0038] f t =δ(W f .[h t-1 ,x t ]+n f )
[0039] i t =δ(W i .[h t-1 ,x t ]+b i )
[0040]
[0041]
[0042] o t =δ(W o [h t-1 ,x t ])+b o
[0043] h t =o t ⊙tanh(c t )
[0044] In the formula, f t represents the forget gate, δ represents the sigmoid function, W f and b f Respectively represent the weight matrix and bias of the forget gate, [h t-1 ,x t ] represents the input vector, h t-1 represents the external state at the previous moment, x t is the input at the current moment; i t Represents the input, W i and b i Represent the weight matrix and bias of the input gate respectively; represents the candidate state, W t and b t Represent their weight matrix and bias respectively, tanh(c t ) is the hyperbolic tangent function; c t represents the internal state of the update unit, c t-1 is the internal state of the unit at the previous moment; o t represents the output gate, W o and b o Represent the weight matrix and bias of the output gate respectively;
[0045] In the process of training the fused comprehensive features based on the LSTM model, mean absolute error, mean absolute percentage error, root mean square error and determination coefficient are used as evaluation indicators for training.
[0046] A lithium-ion battery health status estimation system, comprising:
[0047] The acquisition unit is configured as:
[0048] Used to collect the voltage, current, capacity and corresponding sampling time data of lithium-ion batteries during the constant current charging stage of the charge and discharge cycle;
[0049] The curve generation unit is configured as follows:
[0050] Used to obtain a capacity increment curve based on the voltage, current, capacity and corresponding sampling time data;
[0051] The curve processing unit is configured as follows:
[0052] Used to process the capacity increment curve based on a segmented aggregation approximation algorithm to generate a short sequence capacity increment data curve;
[0053] The image generation unit is configured as follows:
[0054] Used to transform the short sequence capacity increment data curve based on the GAF algorithm to generate image data;
[0055] The fusion unit is configured as:
[0056] Used to extract first dimensional features from the image data using a 2DCNN network, extract second dimensional features from the capacity increment curve using a 1DCNN network, and fuse the first dimensional features and the second dimensional features based on a feature fusion network to generate a fused comprehensive feature;
[0057] Output unit, configured as:
[0058] It is used to train the fused comprehensive features based on the LSTM model and output the health status estimation result.
[0059] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the lithium-ion battery health state estimation method when executing the computer program.
[0060] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the lithium-ion battery health state estimation method are implemented.
[0061] Compared with the prior art, the present invention has the following beneficial technical effects:
[0062] The invention provides a lithium ion battery health state estimation method, system, device and medium, comprising the following steps: collecting voltage, current, capacity and corresponding sampling time data of a lithium ion battery in a constant current charging stage of a charge and discharge cycle; obtaining a capacity increment curve based on the voltage, current, capacity and corresponding sampling time data; processing the capacity increment curve based on a segmented aggregation approximation algorithm to generate a short sequence capacity increment data curve; transforming the short sequence capacity increment data curve based on a GAF algorithm to generate image data; extracting first dimensional features from the image data using a 2DCNN network, extracting second dimensional features from the capacity increment curve using a 1DCNN network, fusing the first dimensional features and the second dimensional features based on a feature fusion network to generate fused comprehensive features; training the fused comprehensive features based on an LSTM model to output a health state estimation result; the present application transforms the short sequence capacity increment data curve to generate image data, while the existing related technology only extracts features related to battery aging from a one-dimensional time series; secondly, the present application adopts a hybrid model combining 2DCNN, 1DCNN and LSTM, and uses the LSTM model to mine and learn fused features to achieve accurate health state estimation. BRIEF DESCRIPTION OF THE DRAWINGS
[0063] Figure 1 A flow chart of a method for estimating the health status of a lithium-ion battery according to an embodiment of the present disclosure is shown;
[0064] Figure 2 is the original IC curve and the normalized IC curve processed by the segmented aggregation approximation algorithm described in this application;
[0065] Figure 3 A process diagram of converting an IC curve into an image using GAF according to an embodiment of the present disclosure is shown;
[0066] Figure 4 A feature fusion network structure diagram of an embodiment of the present disclosure is shown;
[0067] Figure 5 The basic structure of the LSTM model of the embodiment of the present disclosure is shown;
[0068] Figure 6 The prediction results of the embodiments of the present disclosure on NASA lithium-ion battery aging data are shown;
[0069] Figure 7 The prediction results of the embodiments of the present disclosure based on the lithium-ion battery aging data of Oxford University are shown. DETAILED DESCRIPTION
[0070] In the following, only some exemplary embodiments are briefly described. As those skilled in the art will appreciate, the described embodiments may be modified in various ways without departing from the spirit or scope of the present invention. Therefore, the drawings and descriptions are considered to be exemplary and non-restrictive in nature.
[0071] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.
[0072] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.
[0073] It should also be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.
[0074] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.
[0075] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.
[0076] Figure 1 A flow chart of a method for estimating the health status of a lithium-ion battery according to an embodiment of the present disclosure is shown. Figure 1 As shown, the following steps are included:
[0077] Step S1: collecting voltage, current, capacity and corresponding sampling time data of the lithium-ion battery during the constant current charging stage of the charge and discharge cycle;
[0078] Preferably, in step S1 of the embodiment of the present disclosure, the process of obtaining the battery health status, voltage, current, capacity, and sampling time data of each charge and discharge cycle constant current charging stage of the lithium-ion battery is:
[0079] The constant current charging voltage data of the i-th cycle is the sequence U i,0 ,U i,1 ,...,U i,j , where U i,0 and U o,j They are the initial voltage and the final voltage of the constant current charging stage respectively;
[0080] The constant current charging current data of the i-th cycle is sequence I i,0 ,I i,1 ,...,I i,j , where I i,0 and I i,j They are the initial current and the termination current in the constant current charging stage respectively;
[0081] The constant current charging sampling time data of the i-th cycle is the sequence t i,0 ,t i,1 ,...,t i,j , where t i,0 and t i,j They are the initial sampling time and the terminal sampling time of the constant current charging stage respectively;
[0082] The charging capacity data in the constant current charging stage is the sequence Q1, Q2, ..., Q n , Q i =∫I i dt i , where Q i is the constant current charging capacity of the battery in the i-th cycle, I i and t i are the current and sampling time of the i-th cycle respectively.
[0083] Furthermore, in some embodiments, this step also collects health status data of the lithium battery, and the health status data of the lithium battery is used to evaluate the reliability of the final output result of the present application. Specifically, the health status data sequence of the lithium battery is SOH1, SOH2, ..., SOH n ,
[0084] in,
[0085] In the formula, SOH i is the battery health status of the i-th (i=1,2,…,n) charge-discharge cycle, n is the number of battery charge-discharge cycles, C i is the maximum discharge capacity of the battery in the ith cycle, and C is the rated capacity of the battery.
[0086] Step S2: Based on the voltage, current, capacity and corresponding sampling time data, a capacity increment curve is obtained, such as Figure 2 As shown;
[0087] Preferably, in step S2 of the embodiment of the present disclosure, the incremental capacity (IC) curve is:
[0088]
[0089] In the formula, △Q and △U represent the change in capacity and voltage of the battery from time m to time m+1, respectively, and U m+1 represents the battery voltage at time m+1, U m represents the voltage at time m, Q m represents the battery capacity at time m, Q m+1 represents the capacity of the battery at time m+1, I represents the current of the battery, and t m+1 represents the sampling time at time m+1, t m Represents the sampling time at time m.
[0090] Step S3: Processing the capacity increment curve based on a piecewise aggregate approximation (PAA) algorithm to generate a short sequence capacity increment data curve;
[0091] Preferably, in step S3 of the embodiment of the present disclosure, the capacity increment curve is processed based on the segmented aggregation approximation algorithm. It should be noted that, in the embodiment of the present disclosure, the segmented aggregation approximation algorithm simplifies the data volume while retaining the main features and details of the original data;
[0092] The process of processing the capacity increment curve based on the segmented aggregation approximation algorithm to generate a short sequence capacity increment data curve is as follows:
[0093] Given an IC sequence Z1,Z2,...,Z of length m m , turning it into a data sequence of length w where w <m;
[0094] The resulting compression ratio is The dimensionality reduction process of time series data, and satisfy
[0095] In the formula, represents the ith IC data after dimension reduction, Z j represents the jth original IC data, is the compression ratio.
[0096] Step S4: transforming the short sequence capacity increment data curve based on the GAF algorithm to generate image data;
[0097] Preferably, in step S4 of the embodiment of the present disclosure, the process of converting the short sequence capacity increment data curve based on the GAF algorithm to generate the image data is:
[0098] The IC time series data X = {x1, x2, ..., x n}Normalized to the interval [-1,1]:
[0099]
[0100] where x i and represent the original IC sequence and the normalized IC sequence at the i-th time point, respectively; max(X) and min(X) represent the maximum and minimum values of the IC sequence, respectively;
[0101] Convert the IC sequence in Cartesian coordinates to polar coordinates:
[0102]
[0103] Among them, φ i represents the polar angle at the i-th time point, r i Indicates the extreme meridian, t i represents the timestamp, N is the total number of sampling points;
[0104] The GASF coding method is used to encode the IC curve in polar coordinates, and the calculation formula is as follows:
[0105]
[0106] Among them, φ1, φ2, ..., φ n represents the polar angle corresponding to different time points; GASF represents a Gram-like matrix that maintains time dependence.
[0107] It should be noted that the GAF method described in this embodiment can provide feature information and correlation between different vectors at the same time through inner product operations, and can well capture the local and global characteristics of time series data, thereby reflecting more details of battery aging. The GAF method described in the embodiment of the present disclosure is simple to calculate, can avoid complex feature extraction work, and does not require additional filtering of the time series data.
[0108] Step S5: Use the 2DCNN network to extract the first dimension features from the image data, and use the 1DCNN network to extract the second dimension features from the capacity increment curve. The second dimension features refer to the features extracted from the IC time series data. The first dimension features and the second dimension features are fused based on the feature fusion network, such as Figure 4 As shown, a fusion comprehensive feature is generated;
[0109] Preferably, in step S5 of the embodiment of the present disclosure, the process of generating the fusion comprehensive features is:
[0110] The 2DCNN branch is described as:
[0111]
[0112] a 2D =g(y 2D ),a 2D ∈R m
[0113] Among them, W 2D and b 2D denote weight and bias respectively, Represents the input vector, which is a GAF image, a 2D Represents the feature vector after 2DCNN, g(y2D ) represents the activation function.
[0114] The 1DCNN branch is described as:
[0115]
[0116] a 1D =g(y 1D ),a 1D ∈R n
[0117] Among them, W 1D and b 1D denote weight and bias respectively, Represents the input vector, which is an IC sequence, a 1D represents the feature vector after 1DCNN, g(y 1D ) represents the activation function.
[0118] The feature fusion layer is described as:
[0119] x f =concat(a 2D ,a 1D ),x f ∈R m+n
[0120] Among them, concat(a 2D ,a 1D ) represents the feature fusion function, x f Represents the fused feature vector.
[0121] Step S6: Figure 5 As shown, the fused comprehensive features are trained based on the LSTM model to output the health status estimation result;
[0122] Preferably, in step S5 of the embodiment of the present disclosure, the LSTM model is:
[0123] f t =δ(W f .[h t-1 ,x t ]+n f )
[0124] i t =δ(W i .[h t-1 ,x t ]+b i )
[0125]
[0126]
[0127] o t =δ(W o [h t-1 ,x t ])+b o
[0128] h t =o t ⊙tanh(c t )
[0129] In the formula, f t represents the forget gate, δ represents the sigmoid function, W f and b f Respectively represent the weight matrix and bias of the forget gate, [h t-1 ,x t ] represents the input vector, h t-1 represents the external state at the previous moment, x t is the input at the current moment; i t Represents the input, W i and b i Represent the weight matrix and bias of the input gate respectively; represents the candidate state, W t and b t Represent their weight matrix and bias respectively, tanh(c t ) is the hyperbolic tangent function; c t represents the internal state of the update unit, c t-1 is the internal state of the unit at the previous moment; o t represents the output gate, W o and b o Represent the weight matrix and bias of the output gate respectively;
[0130] In the process of training the fused comprehensive features based on the LSTM model, mean absolute error, mean absolute percentage error, root mean square error and determination coefficient are used as evaluation indicators for training;
[0131] Specifically, in the disclosed embodiment, the data in the short sequence capacity increment data curve is a one-dimensional IC sequence, and the image data is a GAF image;
[0132] 40% of the one-dimensional IC sequence and its corresponding GAF image are taken as the training set to train the CNN-LSTM model. 60% of the one-dimensional IC sequence and its corresponding GAF image are taken as the test set and input into the CNN-LSTM model to output the estimated result of the health status.
[0133] 60% of the one-dimensional IC sequence and its corresponding GAF image are taken as the training set to train the CNN-LSTM model. 40% of the one-dimensional IC sequence and its corresponding GAF image are taken as the test set, which are input into the CNN-LSTM model to output the estimated results of the health status.
[0134] 80% of the one-dimensional IC sequence and its corresponding GAF image are taken as the training set to train the CNN-LSTM model. 20% of the one-dimensional IC sequence and its corresponding GAF image are taken as the test set, which are input into the CNN-LSTM model to output the estimated results of the health status.
[0135] In order to demonstrate the feasibility and accuracy of the lithium-ion battery health status estimation method based on GAF and CNN-LSTM in this application, this application uses the following examples to illustrate and compare the MAE, MAPE, RMSE and R of various algorithms. 2 To illustrate the accuracy of this application.
[0136] This application uses the lithium-ion battery cycle life data sets numbered B0005, B0006, B0007, and B0018 of the National Aeronautics and Space Administration (NASA) of a certain country and the lithium-ion battery cycle life data sets numbered Cell 1, Cell 3, Cell 7, and Cell 8 of the University of Oxford to verify the proposed hybrid model. The operating temperature of the two 18650 batteries of NASA is 24°C, the charging cut-off voltage is 4.2V, and the rated capacity is 2Ah. The battery aging experiment is carried out using the constant current constant voltage charging mode and the constant current discharge mode. The rated capacity of the four Kokam lithium-ion batteries of the University of Oxford is 740mAh, the ambient temperature is 40°C, and the constant current constant voltage charging mode and the ARTEMIS urban driving cycle discharge mode are used. Characterization tests are carried out after every 100 cycles. The characterization measurement is 1C constant current charging and discharging, and the data sampling frequency is 1hz.
[0137] Use LSTM model and CNN-LSTM model for comparison
[0138] In order to verify the performance of the model on different training sets, the training set ratios were set to 40%, 60%, and 80%, respectively.
[0139] The SOH prediction results using 60% of the NASA battery training data are as follows Figure 6 As shown, Figure 6 (a) is the health status estimation result of B0005 battery. Figure 6 (b) is the health status estimation result of B0006 battery. Figure 6 (c) is the health status estimation result of B0007 battery. Figure 6(d) is the health status estimation result of B0018 battery. It can be seen that compared with the LSTM and CNN-LSTM models, the prediction performance of the GAF-CNN-LSTM model proposed in the present invention is better, the prediction curve fluctuates less, and the fit to the actual SOH curve is closer. The model also has a good prediction effect in the capacity regeneration area. In contrast, the prediction curves of the LSTM and CNN-LSTM models show obvious fluctuations.
[0140] Average MAE, MAPE, RMSE and R of the three models in different training sets 2 As shown in Table 1. It can be seen that for B0005 battery, the MAE, MAPE, RMSE and R 2 The MAE, MAPE, RMSE and R of the CNN-LSTM model are 0.0210, 0.0219, 0.0194 and 0.6021 respectively. 2 The MAE, MAPE, RMSE and R of the GAF-CNN-LSTM model are 0.0150, 0.0170, 0.0169 and 0.7858 respectively. 2 The MAE, MAPE, and RMSE of the GAF-CNN-LSTM model are smaller than those of the LSTM model and the CNN-LSTM model. 2 Higher than the LSTM model and CNN-LSTM model.
[0141] Table 2 shows the average error indicators of the four batteries B0005, B0006, B0007, and B0018. The MAE, MAPE, and RMSE of the GAF-CNN-LSTM model are 0.0027, 0.0038, and 0.0033, respectively, all of which do not exceed 0.004. Compared with the LSTM model, the MAE, MAPE, and RMSE of the GAF-CNN-LSTM model are reduced by 83.8%, 82.6%, and 83.0%, respectively. Compared with the CNN-LSTM model, the average MAE, MAPE, and RMSE of the GAF-CNN-LSTM model are reduced by 77.1%, 74.3%, and 76.3%, respectively. The R 2 is 0.9890, the highest among the three models, 45.6% and 16.9% higher than the LSTM and CNN-LSTM models, respectively. In summary, the GAF-CNN-LSTM model proposed in this paper shows good performance in NASA's lithium-ion battery aging dataset. Its MAE, MAPE, RMSE, R 2 They are better than the LSTM model and the CNN-LSTM model, which shows that they have better stability and accuracy.
[0142] Table 1 Error indicators of model comparison experiments
[0143]
[0144]
[0145] Table 2 Average error indicators of four batteries
[0146]
[0147]
[0148] The SOH prediction results using 60% of the Oxford University battery training data are as follows: Figure 7 As shown, Figure 7 (a) is the estimated result of the health status of Cell 1. Figure 7 (b) is the estimated result of the health status of Cell 3 battery. Figure 7 (c) is the estimated result of the health status of Cell 7 battery. Figure 7 (d) is the health status estimation result of Cell 8. It can be seen that compared with the LSTM and CNN-LSTM models, the GAF-CNN-LSTM model proposed in the present invention has better prediction effect, smaller fluctuation of the prediction curve, and is closer to the actual SOH curve.
[0149] Average MAE, MAPE, RMSE and R of the three models in different training sets 2 As shown in Table 3. It can be seen that for Cell1 battery, the MAE, MAPE, RMSE and R 2 The MAE, MAPE, RMSE and R of the CNN-LSTM model are 0.0164, 0.0210, 0.0172 and 0.6084 respectively. 2 The MAE, MAPE, RMSE and R of the GAF-CNN-LSTM model are 0.0108, 0.0138, 0.0125 and 0.6975 respectively. 2 The MAE, MAPE, and RMSE of the GAF-CNN-LSTM model are smaller than those of the LSTM model and the CNN-LSTM model. 2 Higher than the LSTM model and CNN-LSTM model.
[0150] Table 4 shows the average error indicators of the four cells Cell 1, Cell 3, Cell 7, and Cell 8. The MAE, MAPE, and RMSE of the GAF-CNN-LSTM model are 0.0017, 0.0022, and 0.0021, respectively, all within 0.003. Compared with the LSTM model, the MAE, MAPE, and RMSE of the GAF-CNN-LSTM model are reduced by 88.4%, 88.3%, and 87.0%, respectively. Compared with the CNN-LSTM model, the MAE, MAPE, and RMSE of the GAF-CNN-LSTM model are reduced by 88.1%, 88.0%, and 86.9%, respectively. The R 2 is 0.9881, which is 55.9% and 58.6% higher than the LSTM and CNN-LSTM models, respectively. In summary, the GAF-CNN-LSTM model proposed in the present invention shows good accuracy and stability in the lithium-ion battery aging data of Oxford University, and its prediction accuracy and fitting effect are better than those of the LSTM and CNN-LSTM models.
[0151] Table 3 Error indicators of model comparison experiments
[0152]
[0153]
[0154] Table 4 Average error indicators of four batteries
[0155]
[0156] The present disclosure also provides a lithium-ion battery health status estimation system, including:
[0157] The acquisition unit is configured as:
[0158] Used to collect the voltage, current, capacity and corresponding sampling time data of lithium-ion batteries during the constant current charging stage of the charge and discharge cycle;
[0159] The curve generation unit is configured as follows:
[0160] Used to obtain a capacity increment curve based on the voltage, current, capacity and corresponding sampling time data;
[0161] The curve processing unit is configured as follows:
[0162] Used to process the capacity increment curve based on a segmented aggregation approximation algorithm to generate a short sequence capacity increment data curve;
[0163] The image generation unit is configured as follows:
[0164] Used to transform the short sequence capacity increment data curve based on the GAF algorithm to generate image data;
[0165] The fusion unit is configured as:
[0166] Used to extract first dimensional features from the image data using a 2DCNN network, extract second dimensional features from the capacity increment curve using a 1DCNN network, and fuse the first dimensional features and the second dimensional features based on a feature fusion network to generate a fused comprehensive feature;
[0167] Output unit, configured as:
[0168] It is used to train the fused comprehensive features based on the LSTM model and output the health status estimation result.
[0169] In another embodiment of the present invention, a computer device is provided, the computer device including a processor and a memory, the memory is used to store a computer program, 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., which are the computing core and control core of the terminal, which are suitable for implementing one or more instructions, and are 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 lithium-ion battery health state estimation method.
[0170] In another embodiment of the present invention, the present invention 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 can be understood that the computer-readable storage medium here can include both built-in storage media in the computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by the 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 lithium-ion battery health state estimation method described in the above embodiment.
[0171] 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.
[0172] 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.
[0173] 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.
[0174] 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.
[0175] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive. The scope of the present invention is defined by the attached claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any figure mark in the claims should not be regarded as limiting the claims involved.
[0176] In addition, it should be understood that although this specification is described in accordance with the implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation modes that can be understood by those skilled in the art. The above content is only to illustrate the technical idea of the present invention, and cannot be used to limit the protection scope of the present invention. Any changes made on the basis of the technical solution according to the technical idea proposed by the present invention shall fall within the protection scope of the claims of the present invention.
Claims
1. A method for estimating the health status of a lithium-ion battery, characterized in that: The following steps are involved: Collect the voltage, current, capacity and corresponding sampling time data of lithium-ion batteries during the constant current charging stage of the charge and discharge cycle; Obtaining a capacity increment curve based on the voltage, current, capacity and corresponding sampling time data; Processing the capacity increment curve based on a segmented aggregation approximation algorithm to generate a short sequence capacity increment data curve; The short sequence capacity increment data curve is transformed based on the GAF algorithm to generate image data; A 2DCNN network is used to extract first dimensional features from the image data, a 1DCNN network is used to extract second dimensional features from the capacity increment curve, and the first dimensional features and the second dimensional features are fused based on a feature fusion network to generate fused comprehensive features; The fused comprehensive features are trained based on the LSTM model to output the health status estimation result.
2. The method for estimating the health status of a lithium-ion battery according to claim 1, characterized in that: The process of obtaining the battery health status, voltage, current, capacity, and sampling time data of each charge and discharge cycle of the lithium-ion battery at the constant current charging stage is as follows: The constant current charging voltage data of the i-th cycle is the sequence U i,0 ,U i,1 ,...,U i,j , where U j,0 and U i,j They are the initial voltage and the final voltage of the constant current charging stage respectively; The constant current charging current data of the i-th cycle is sequence I i,0 ,I i,1 ,...,I i,j , where I i,0 and I i,j They are the initial current and the termination current in the constant current charging stage respectively; The constant current charging sampling time data of the i-th cycle is the sequence t i,0 ,t i,1 ,...,t i,j , where t i,0 and t i,j They are the initial sampling time and the terminal sampling time of the constant current charging stage respectively; The charging capacity data in the constant current charging stage is the sequence Q1, Q2, ..., Q n , Q i =∫I i dt i , where Q i is the constant current charging capacity of the battery in the i-th cycle, I i and t i are the current and sampling time of the i-th cycle respectively.
3. The method for estimating the health status of a lithium-ion battery according to claim 1, characterized in that: The capacity increment curve is: In the formula, △Q and △U represent the change in capacity and voltage of the battery from time m to time m+1, respectively, and U m+1 represents the battery voltage at time m+1, U m represents the voltage at time m, Q m represents the battery capacity at time m, Q m+1 represents the capacity of the battery at time m+1, I represents the current of the battery, and t m+1 represents the sampling time at time m+1, t m Represents the sampling time at time m.
4. The method for estimating the health status of a lithium-ion battery according to claim 1, characterized in that: The process of processing the capacity increment curve based on the segmented aggregation approximation algorithm to generate a short sequence capacity increment data curve is as follows: Given an IC sequence Z1,Z2,...,Z of length m m , turning it into a data sequence of length w where w <m; The resulting compression ratio is The dimensionality reduction process of time series data, and satisfy In the formula, represents the ith IC data after dimension reduction, Z j represents the jth original IC data, is the compression ratio.
5. The method for estimating the health status of a lithium-ion battery according to claim 1, characterized in that: The process of converting the short sequence capacity increment data curve based on the GAF algorithm to generate image data is as follows: Among them, x i and represent the original IC sequence and the normalized IC sequence at the i-th time point, max(X) and min(X) represent the maximum and minimum values of the IC sequence, respectively; φ i represents the polar angle at the i-th time point, r i Indicates the extreme meridian, t i represents the timestamp, N is the total number of sampling points; GASF represents a Gram-like matrix that maintains time dependency.
6. The method for estimating the health status of a lithium-ion battery according to claim 1, characterized in that: The process of generating fusion comprehensive features is as follows: to 2D =g(y 2D ),to 2D ∈R m to 1D =g(y 1D ),to 1D ∈R n x f =concat(a 2D ,a 1D ),x f ∈R m+n Where W 2D and W 1D Represent the weight matrices of 2DCNN and 1DCNN respectively, a 2D and a 1D Respectively represent the feature vectors after 2DCNN and 1DCNN; g(y 1D ), g(y 2D ) represents the activation function, concat(a 2D ,a 1D ) represents the feature fusion function, x f Represents the fused feature vector.
7. The method for estimating the health status of a lithium-ion battery according to claim 1, characterized in that: The LSTM model is: f t =δ(W f .[h t-1 ,x t ]+b f ) i t =δ(W i ·[h t-1 ,x t ]+b i ) o t =δ(W o [h t-1 ,x t ])+b o h t =o t ⊙ttanh(c t ) In the formula, f t represents the forget gate, δ represents the sigmoid function, W f and b f Respectively represent the weight matrix and bias of the forget gate, [h t-1 ,x t ] represents the input vector, h t-1 represents the external state at the previous moment, x t is the input at the current moment; i t Represents the input, W i and b i Represent the weight matrix and bias of the input gate respectively; represents the candidate state, W t and b t Represent their weight matrix and bias respectively, tanh(c t ) is the hyperbolic tangent function; c t represents the internal state of the update unit, c t-1 is the internal state of the unit at the previous moment; o t represents the output gate, w o and b o Represent the weight matrix and bias of the output gate respectively; In the process of training the fused comprehensive features based on the LSTM model, mean absolute error, mean absolute percentage error, root mean square error and determination coefficient are used as evaluation indicators for training.
8. A lithium-ion battery health status estimation system, characterized in that: The method for estimating the health status of a lithium-ion battery according to any one of claims 1 to 7 comprises: The acquisition unit is configured as: Used to collect the voltage, current, capacity and corresponding sampling time data of lithium-ion batteries during the constant current charging stage of the charge and discharge cycle; The curve generation unit is configured as follows: Used to obtain a capacity increment curve based on the voltage, current, capacity and corresponding sampling time data; The curve processing unit is configured as follows: Used to process the capacity increment curve based on a segmented aggregation approximation algorithm to generate a short sequence capacity increment data curve; The image generation unit is configured as follows: Used to transform the short sequence capacity increment data curve based on the GAF algorithm to generate image data; The fusion unit is configured as: Used to extract first dimensional features from the image data using a 2DCNN network, extract second dimensional features from the capacity increment curve using a 1DCNN network, and fuse the first dimensional features and the second dimensional features based on a feature fusion network to generate a fused comprehensive feature; Output unit, configured as: It is used to train the fused comprehensive features based on the LSTM model and output the health status estimation result.
9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the lithium-ion battery health state estimation method according to any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the lithium-ion battery health state estimation method as described in any one of claims 1 to 7 are implemented.
Citation Information
Cited By
Lithium ion battery health state evaluation method and device based on image recognition
CN119986382A
Lithium ion battery SOH estimation system and method based on automatic neural architecture search
CN120686089A
