Lithium ion battery capacity increment curve reconstruction method under incomplete charging and discharging
The capacity increment curve of lithium-ion batteries is reconstructed through the parameterless model and maximum likelihood estimation method, which solves the accuracy of battery health status evaluation under incomplete charge and discharge data, and improves battery safety and service life.
Patent Information
- Application Number
- CN202410915539.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-09
- Publication Date
- 2024-11-08
AI Technical Summary
The prior art is difficult to effectively reconstruct the capacity increment curve of lithium-ion batteries through incomplete charge and discharge data, resulting in inaccurate evaluation of battery health status and affecting battery safety and service life.
Using parameterless model construction and maximum likelihood estimation method, differential calculation and Gaussian filtering smooth preprocessing data, combined with function principal component analysis, incomplete lithium-ion battery capacity increment curve is reconstructed, and online evaluation of battery status is provided. in accordance with.
The effective reconstruction of the capacity increment curve of the lithium-ion battery under incomplete charge and discharge conditions is achieved, the reliability and safety of battery status evaluation is improved, and the disadvantages of traditional methods relying on complete charge and discharge data are overcome.
Smart Images

Figure CN118914850A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of lithium-ion battery state estimation, and in particular relates to a method for reconstructing a capacity increment curve of a lithium-ion battery under incomplete charging and discharging. Background Art
[0002] Whether the battery management system (BMS) can effectively intervene and manage the status of lithium-ion batteries to ensure safe and reliable operation of the battery depends on the accurate estimation of the battery status, especially the battery's state of health (SOH). SOH is characterized by the battery's capacity and internal resistance and is an internal parameter of the battery. In actual applications, BMS can only obtain online operating parameters such as battery current, voltage, and temperature through sensor units. Therefore, how to use these online data to reveal the degradation behavior of the battery is particularly important.
[0003] By using the battery online operating parameters, the incremental capacity (IC) curve of battery charging and discharging can be obtained. The complete IC curve can reflect the changes in the phase change reactions of the active substances and electrode materials inside the battery, and can provide more comprehensive battery degradation information. Existing literature ([1] D. Anseán, VM García, M. González, C. Blanco-Viejo, JC Viera, YF Pulido, et al. Lithium-ion battery degradation indicators via incremental capacity analysis. IEEE Transactions on Industry Applications. 2019; 55: 2992-3002. [2] Z. Zhang, H. Min, H. Guo, Y. Yu, W. Sun, J. Jiang, et al. State of health estimation method for lithium-ion batteries using incremental capacity and long short-term memory network. Journal of Energy Storage. 2023; 64: 107063.) mainly uses voltage, current and other data during the complete charge and discharge process of the battery to obtain IC curves to reveal the condition of the battery; however, in actual applications, the battery is rarely fully charged and discharged, and only fragmentary IC curves can be obtained.
[0004] In order to effectively evaluate the SOH of the battery, the complete IC curve needs to be reconstructed. This process not only helps to reveal the actual degradation of the battery, but also provides more accurate SOH evaluation data for the BMS, thereby improving the safety and service life of the battery. There is currently no relevant literature published. Summary of the invention
[0005] In order to overcome the above-mentioned difficulties, the present invention provides a method for reconstructing the capacity increment curve of a lithium-ion battery under incomplete charge and discharge, which provides a reliable basis for online evaluation of battery SOH, thereby improving the safety and service life of the battery.
[0006] In order to achieve the above object, the technical solution adopted by the present invention is:
[0007] A method for reconstructing a capacity increment curve of a lithium-ion battery under incomplete charge and discharge comprises obtaining a complete IC curve offline, constructing a non-parameterized model and reconstructing an online segment IC curve; obtaining a complete IC curve offline is to perform differential calculation and Gaussian filtering and smoothing preprocessing on voltage and current data collected by a battery management system (BMS) to obtain a complete IC curve under different aging conditions; constructing a non-parameterized model is to construct a non-parameterized model using a function principal component analysis (FPCA); reconstructing an online segment IC curve is to calculate the principal component score using a maximum likelihood estimation method to reconstruct the segment IC curve under online incomplete charge and discharge.
[0008] A method for reconstructing a capacity increment curve of a lithium-ion battery under incomplete charge and discharge conditions comprises the following steps:
[0009] 1) Obtain complete IC curve offline:
[0010] The complete charging online parameters of the battery under different aging conditions are obtained through BMS, and the complete IC curves under different aging conditions are obtained by using the data preprocessing method of differential calculation and smoothing filtering. The specific steps are as follows:
[0011] 1.1) Get battery online parameters:
[0012] Obtain complete battery charging data through BMS, including current data (I n,k,1 , ..., I n,k,i , ..., I n,k,l )′, voltage data (V n,k,1 , ..., V n,k,i , .., V n,k,l )′, and its corresponding time series is (T n,k,1 , ..., T n,k,i , ..., T n,k,l )′, where i∈1,...,l represents the ith data point of the kth charging cycle of the nth battery; the charging capacity sequence Qn,k,i for:
[0013]
[0014] 1.2) Differential calculation:
[0015] By differentiating the charge capacity with respect to the voltage, IC is calculated:
[0016]
[0017] Among them, IC n,k,i represents the ith IC value of the kth charging cycle of the nth battery;
[0018] 1.3) Gaussian filter smoothing:
[0019] The Gaussian filter function is used to smooth the IC data. The smoothed IC data is:
[0020]
[0021] in, is a Gaussian function, σ is the standard deviation;
[0022] 2) Non-parametric model construction:
[0023] The IC curve reconstruction model was constructed using the preprocessed IC data; the non-parametric modeling method of functional principal component analysis was adopted, and the general formula of the model was:
[0024] IC n,k (v) = μ(v) + x n,k (v)+ω n,k (v) (4)
[0025] Among them, IC n,k (v) represents the IC value of the nth battery at the kth charging cycle when the voltage is v; μ(v) represents the mean function of the overall IC curve; x n,k (v) represents the deviation of the kth charging cycle of the nth battery from the mean function, and the covariance G(v, v′)=cov(x n,k (v), x n,k (v′)), v′ is the transpose of v; ω n,k (v) has a mean of 0 and a variance of Noise term; According to the Karhunen-Loéve theory, G(v, v′) is expanded as:
[0026]
[0027] Among them, ψ p(v) is the pth functional principal component (FPC), p∈1,2,...,λ p is the corresponding principal component eigenvalue, x n,k (v) Projection onto a series of basis functions:
[0028]
[0029] ξ n,k,p It represents the pth principal component score corresponding to the kth charging cycle of the nth battery; the IC curve of the kth charging cycle of the nth battery is approximately expressed as:
[0030]
[0031] The mean function of the model is calculated using the local linear smoothing method; the covariance
[0032]
[0033] The expected covariance E(G n,k (v n,k,i , v n,k,j ))for:
[0034]
[0035] When estimating the covariance matrix surface, the diagonal lines containing noise are removed and the local linear smoothing method is used to calculate and a diagonal function with noise terms Noise term Among them, v max , v0 are the upper and lower limits of the battery charging cut-off voltage respectively;
[0036] Characteristic function of the covariance matrix and the corresponding eigenvalues The following equations are satisfied:
[0037]
[0038] Among them, p∈1, 2, ..., represents the pth characteristic function, that is, the principal component of the function; Represents the eigenvalue corresponding to the characteristic function;
[0039] 3) Online fragment IC curve reconstruction:
[0040] Use BMS to get the online segment IC curve IC n,k =(IC n,k,1 ,…,IC n,k,l )′, the corresponding voltage sequence is V n,k =(V n,k,1 , …, V n,k,l)′, the likelihood function is constructed using the maximum likelihood estimation method
[0041]
[0042] in, represents the mean function of the kth charging cycle of the nth battery; The principal component of the function representing the kth charging cycle of the nth battery, is the pth characteristic function; ξ n,k represents the score of the principal component of the kth charging cycle function of the nth battery;
[0043] make Calculate the function principal component score
[0044]
[0045] Online IC reconstruction curve for:
[0046]
[0047] Online IC reconstruction curve The 100(1-α)% confidence interval for is:
[0048]
[0049] in, is a standard normal distribution Quantile.
[0050] Compared with the prior art, the present invention has the following beneficial effects:
[0051] The present invention proposes a method for reconstructing the capacity increment curve of a lithium-ion battery under incomplete charging and discharging, constructs a battery charging IC curve model by using a parameter-free modeling method, and reconstructs the online fragment IC curve using a maximum likelihood estimation algorithm. The present invention realizes the reconstruction of the complete IC curve of the battery under incomplete charging and discharging, provides a reliable basis for online evaluation of the battery SOH, and overcomes the disadvantage that the traditional method needs to rely on the complete charging and discharging data of the battery. BRIEF DESCRIPTION OF THE DRAWINGS
[0052] Figure 1 It is a flow chart of the present invention.
[0053] Figure 2 The box plot is a root mean square error of the reconstructed IC curve of the lithium-ion battery under different missing rates in the embodiment.
[0054] Figure 3It is a comparison diagram of the IC curve reconstructed by the embodiment and the actual curve. DETAILED DESCRIPTION
[0055] The present invention is further described in detail below with reference to the accompanying drawings and embodiments.
[0056] Reference Figure 1 , a method for reconstructing a capacity increment curve of a lithium-ion battery under incomplete charge and discharge, comprising the following steps:
[0057] 1) Obtain the complete IC curve under a specific charging protocol offline:
[0058] Under a specific charging scheme, the complete charging online parameters of the battery under different aging conditions are obtained through the BMS, and the complete IC curves under different aging conditions are obtained by using data preprocessing methods such as differential calculation and smoothing filtering. The specific steps are as follows:
[0059] 1.1) Get battery online parameters:
[0060] Obtain complete battery charging data through BMS, mainly including current data (I n,k,1 , ..., I n,k,i , ..., I n,k,l )′, voltage data (V n,k,1 , ..., V n,k,i , ..., V n,k,l )′, and its corresponding time series is (T n,k,1 , ..., T n,k,i , ..., T n,k,l )′, where i∈1,...,l represents the i-th data point of the k-th charging cycle of the n-th battery. The charging capacity sequence Q n,k,i for:
[0061]
[0062] 1.2) Differential calculation:
[0063] The IC value is calculated by differentiating the charge capacity with respect to the voltage:
[0064]
[0065] Among them, IC n,k,i represents the ith IC value of the kth charging cycle of the nth battery;
[0066] 1.3) Gaussian filter smoothing:
[0067] The Gaussian filter function is used to smooth the IC data, reduce the impact of sampling noise on the differential result, and retain key information such as IC peak value to the greatest extent. The smoothed IC data is
[0068]
[0069] in, is a Gaussian function, σ is the standard deviation, which is used to control the width of the Gaussian function. The larger the standard deviation, the smoother the filtering effect;
[0070] 2) Non-parametric model construction:
[0071] The IC curve reconstruction model was constructed using the preprocessed IC data; the non-parametric modeling method of functional principal component analysis was adopted, and the general formula of the model was:
[0072] IC n,k (v) = μ(v) + x n,k (v)+ω n,k (v) (4)
[0073] Among them, IC n,k (v) represents the IC value of the nth battery at the kth charging cycle when the voltage is v; μ(v) represents the mean function of the overall IC curve; x n,k (v) represents the deviation of the kth charging cycle of the nth battery from the mean function, which has a random characteristic, its mean is 0, and the covariance G(v, v') = cov(x n,k (v), x n,k (v')), v' is the transpose of v; ω n,k (v) has a mean of 0 and a variance of Noise term; According to the Karhunen-Loéve theory, G(v, v') can be expanded as:
[0074]
[0075] Among them, ψ p (v) is the pth functional principal component (FPC), p∈1,2,…,λ p is the corresponding principal component eigenvalue, x n,k (v) can be projected onto a series of basis functions:
[0076]
[0077] ξ n,k,p It represents the pth principal component score corresponding to the kth charging cycle of the nth battery; the IC curve of the kth charging cycle of the nth battery can be approximately expressed as:
[0078]
[0079] The mean function of the model can be calculated using the local linear smoothing method; the covariance
[0080]
[0081] The expected covariance E(G n,k (v n,k,i , v n,k,j ))for:
[0082]
[0083] Among them, when i = j, δ i,j is equal to 1, and 0 in other cases. Therefore, when estimating the covariance matrix surface, it is necessary to remove the diagonal lines containing noise and use the local linear smoothing method to calculate and a diagonal function with noise terms Noise term Among them, v max , v0 are the upper and lower limits of the battery charging cut-off voltage respectively;
[0084] Characteristic function of the covariance matrix and the corresponding eigenvalues The following equations are satisfied:
[0085]
[0086] Among them, p∈1, 2, ..., represents the pth characteristic function, that is, the principal component of the function; Indicates the eigenvalue corresponding to the eigenfunction; selects the first p eigenfunctions whose cumulative eigenvalues exceed 95%;
[0087] 3) Online fragment IC curve reconstruction:
[0088] Use BMS to get the online segment IC curve IC n,k =(IC n,k,1 ,…,IC n,k,l )′, the corresponding voltage sequence is V n,k =(V n,k,1 , …, V n,k,l )′, the likelihood function is constructed using the maximum likelihood estimation method
[0089]
[0090] in, represents the mean function of the kth charging cycle of the nth battery; The principal component of the function representing the kth charging cycle of the nth battery, is the pth characteristic function; ξ n,k represents the score of the principal component of the kth charging cycle function of the nth battery;
[0091] make Calculate the function principal component score
[0092]
[0093] Online IC reconstruction curve for:
[0094]
[0095] Online IC reconstruction curve The 100(1-α)% confidence interval for is:
[0096]
[0097] in, is a standard normal distribution Quantile.
[0098] This example is based on a nickel cobalt manganese oxide battery degradation data set to verify the effectiveness of the present invention: the data set is released by Xi'an Jiaotong University, the battery has a nominal capacity of 2Ah, a nominal voltage of 3.6V, a charge cut-off voltage and a discharge cut-off voltage of 4.2V and 2.5V, respectively, and the entire experiment is conducted at room temperature; 8 batteries from batch 1 are selected for k-fold cross validation, and the data missing rate is set to 10% to 50% for a total of five groups; the root mean square error (RMSE) indicator is used to measure the effect of the IC curve reconstruction of the present invention; Figure 2 As shown, the mean IC curve reconstruction error is less than 0.25 at different missing rates, indicating that this embodiment can effectively reconstruct a complete IC curve; Figure 3 The comparison between the IC curve reconstructed by this embodiment and the actual curve is shown. It can be seen that the reconstructed curve fits the actual curve data closely, and the actual curve data points are all within the 95% confidence interval of the reconstructed curve, indicating that the method of this embodiment can achieve higher accuracy.
[0099] The present invention is suitable for online battery SOH evaluation and can realize reconstruction of a complete IC curve under different incomplete charge and discharge conditions, thereby improving the reliability of online evaluation of battery SOH and making the model construction more flexible.
Claims
1. A method for reconstructing a capacity increment curve of a lithium-ion battery under incomplete charge and discharge, characterized in that: It includes offline acquisition of complete IC curve, construction of non-parameterized model and reconstruction of online fragmentary IC curve; offline acquisition of complete IC curve is to pre-process the voltage and current data collected by the battery management system (BMS) by differential calculation and Gaussian filtering to obtain complete IC curve under different aging conditions; non-parameterized model construction is to construct a non-parameterized model using functional principal component analysis (FPCA); online fragmentary IC curve reconstruction is to calculate the principal component score using the maximum likelihood estimation method to reconstruct the fragmentary IC curve under incomplete online charging and discharging.
2. A method for reconstructing a capacity increment curve of a lithium-ion battery under incomplete charge and discharge, characterized in that: The following steps are involved: 1) Obtain complete IC curve offline: The complete charging online parameters of the battery under different aging conditions are obtained through BMS, and the complete IC curves under different aging conditions are obtained by using the data preprocessing method of differential calculation and smoothing filtering. The specific steps are as follows: 1.1) Get battery online parameters: Obtain complete battery charging data through BMS, including current data (I n,k,1 ,…,I n,k,i ,…,I n,k,l )′, voltage data (V n,k,1 ,…,V n,k,i ,…,V n,k,l )′, and its corresponding time series is (T n,k,1 ,…,T n,k,i ,…,T n,k,l )′, where i∈1,…,l represents the i-th data point of the k-th charging cycle of the n-th battery; the charging capacity sequence Q n,k,i for: 1.2) Differential calculation: By differentiating the charge capacity with respect to the voltage, IC is calculated: Among them, IC n,k,i represents the ith IC value of the kth charging cycle of the nth battery; 1.3) Gaussian filter smoothing: The Gaussian filter function is used to smooth the IC data. The smoothed IC data is: in, is a Gaussian function, σ is the standard deviation; 2) Non-parametric model construction: The IC curve reconstruction model was constructed using the preprocessed IC data; the non-parametric modeling method of functional principal component analysis was adopted, and the general formula of the model was: IC n,k (v)=μ(v)+x n,k (v)+ω n,k (v) (4) Among them, IC n,k (v) represents the IC value of the nth battery at the kth charging cycle when the voltage is v; μ(v) represents the mean function of the overall IC curve; x n,k (v) represents the deviation of the kth charging cycle of the nth battery from the mean function, and the covariance G(v,v') = cov(x n,k (v),x n,k (v')), v' is the transpose of v; ω n,k (v) has a mean of 0 and a variance of Noise term; According to the Karhunen-Loéve theory, G(v,v') is expanded as: Among them, ψ p (v) is the pth functional principal component (FPC), p∈1,2,…,λ p is the corresponding principal component eigenvalue, x n,k (v) Projection onto a series of basis functions: ξ n,k,p It represents the pth principal component score corresponding to the kth charging cycle of the nth battery; the IC curve of the kth charging cycle of the nth battery is approximately expressed as: The mean function of the model is calculated using the local linear smoothing method; the covariance The expected covariance E(G n,k (v n,k,i ,v n,k,j ))for: When estimating the covariance matrix surface, the diagonal lines containing noise are removed and the local linear smoothing method is used to calculate and a diagonal function with noise terms Noise term Among them, v max , v0 are the upper and lower limits of the battery charging cut-off voltage respectively; Characteristic function of the covariance matrix and the corresponding eigenvalues The following equations are satisfied: Among them, p∈1,2,…, represents the pth characteristic function, that is, the principal component of the function; Represents the eigenvalue corresponding to the characteristic function; 3) Online fragment IC curve reconstruction: Use BMS to get the online segment IC curve IC n,k =(IC n,k,1 ,…,IC n,k,l )′, and its corresponding voltage sequence is V n,k =(V n,k,1 ,…,V n,k,l )′, the likelihood function is constructed using the maximum likelihood estimation method in, represents the mean function of the kth charging cycle of the nth battery; The principal component of the function representing the kth charging cycle of the nth battery, is the pth characteristic function; ξ n,k represents the score of the principal component of the kth charging cycle function of the nth battery; make Calculate the function principal component score Online IC reconstruction curve for: Online IC reconstruction curve The 100(1-α)% confidence interval for is: in, is a standard normal distribution Quantile.
Citation Information
Cited By
Lithium battery health state estimation method and system based on partial charging curve reconstruction
CN119644153A
Energy storage battery health state evaluation method and device in random charging scene, equipment and storage medium
CN119805250A
