Battery state of health intelligent prediction method based on random quantization data enhancement

Abstract

A battery state of health intelligent prediction method based on random quantization data enhancement first obtains existing battery charging data through a BMS, and performs linear interpolation downsampling operation on the data to unify the length, and then constructs an original training set and a test set; subsequently, a random quantizer is constructed, and the original training set data is randomly quantized using the random quantizer to generate enhanced data, and the obtained enhanced data is combined with the original training set data to form an enhanced training set; then, the data sets before and after enhancement are respectively input into an SOH prediction model, the average values of the test losses of the two are respectively calculated and compared, if the test loss after enhancement is smaller than the loss before enhancement, the effectiveness of the battery state of health intelligent prediction method based on random quantization data enhancement is verified; the method can effectively solve the SOH estimation problem under the condition of limited battery data.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of battery state estimation, and particularly relates to a battery health state intelligent prediction method based on random quantization data enhancement. BACKGROUND

[0002] Whether a battery management system (BMS) can effectively intervene and manage the battery state to ensure safe and reliable operation of the battery depends on the accurate estimation of the battery health state (SOH). The SOH is characterized by the capacity and internal resistance of the battery and is an internal parameter of the battery. Existing SOH estimation methods can be generally divided into: measurement-based methods, model-based methods, and data-driven methods ([1] Hu X, Xu L, Lin X, et al. Battery lifetime prognostics [J]. Joule, 2020, 4(2): 310-346.). The measurement-based method relies on direct measurement of the capacity and internal resistance of the battery, and the test conditions are harsh, the test period is long, and it cannot be applied to online evaluation; the model-based method mainly estimates the SOH by establishing a physical model of the battery, which can better describe the internal characteristics of the battery, but is limited by the complexity and nonlinear behavior of the model, and the result accuracy may decrease; the data-driven method mainly uses deep learning, support vector machines and other methods to estimate the SOH, and needs to use a large amount of battery data for training to achieve accurate estimation of the SOH. Compared with direct measurement and physical model methods, the data-driven method can achieve efficient and accurate estimation of the SOH without detailed understanding of the internal mechanism of the battery through big data analysis and training, so the research on the data-driven method for battery SOH is increasing.

[0003] Although the data-driven method can achieve high accuracy through big data training, in actual application, the data of some battery models may be limited and cannot meet the training needs of deep learning or other data-driven algorithms. Therefore, how to achieve accurate SOH estimation under the condition of limited battery data has become a key challenge. SUMMARY

[0004] In order to overcome the above-mentioned shortcomings of the prior art, the purpose of the present application is to provide a battery health state intelligent prediction method based on random quantization data enhancement, which increases the diversity and quantity of the data set through data enhancement technology, improves the generalization ability of the model, and effectively improves the prediction accuracy of the SOH under the condition of limited battery data.

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

[0006] A battery health state intelligent prediction method based on random quantization data enhancement, comprising the following steps:

[0007] Step 1: Obtain existing battery charging data via BMS, including voltage data (V). n，k，1 , ..., V n，k,i , ..., V n，k,l ), current data (I n，k，1 , ..., I n，k，i , ..., I 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 in the k-th charging cycle of the n-th battery;

[0008] Step 2: Perform linear interpolation downsampling on the existing battery charging data obtained in Step 1, downsampling all battery voltage data, current data, and time series data to the same data length l′;

[0009]

[0010]

[0011] Among them, i p ′ represents the location of the sampling point, V n，k，p I n，k，p T n，k，p V represents the value of the p-th sampling point in the k-th cycle of the n-th battery after downsampling. n，k，i I n，k， i, T n，k，i This represents the value of the i-th value of the n-th battery in the k-th cycle before downsampling, where I is the value obtained by rounding down ip', and I+1 is the value of i. p The value obtained by rounding up is...

[0012] Step 3: Construct the voltage, current, and time sample dataset O after linear interpolation downsampling. n ={O n，1 O n，2 O n，3 , ...O n， k, ...O n，K}, where O n，k O represents the k-th cycle of the n-th battery. n，k =(V n，k I n，k T n，k That is, each sample contains three channels, and the three channels store voltage data V respectively. n，k and current data I n，k and time data Tn,k V n,k I n，k T n，k Each set consists of l′ data points; and the original training set Train and the test set Test are divided. One battery data point is selected as the test set Test, and the other battery data points are used for the original training set Train.

[0013] Step 4: Construct a random quantizer. The random quantizer dynamically quantizes the data in each channel, and the quantized values ​​are randomly sampled from randomly divided intervals.

[0014] Step 5: Determine the number of segmentation intervals m, and use a random quantizer to randomly quantize the original training set Train. The voltage, current, and time data in each sample are effectively preserved, generating enhanced samples and thus the generated training set Train′. Then, merge the original training set Train and the randomly quantized enhanced training set Train′ to obtain the enhanced training set Train. aug =Train∪Train′;

[0015] Step 6: Construct a battery state of health (SOH) prediction model. The battery state of health (SOH) prediction model is based on the ResNet network architecture, including more than three residual blocks and a predictor. Each residual block contains two convolutional layers, and batch normalization and ReLU activation functions are applied after the convolutional layers. Finally, the predictor with two fully connected layers maps the features extracted from the residual blocks to a scalar, which is used to output the battery state of health (SOH).

[0016] Step 7: Connect the original training set (Train) and the enhanced training set (Train) separately. aug =Train∪Train′ Input the battery state of health (SOH) prediction model into each of the following methods, obtain the SOH prediction results, and calculate the loss between the predicted and actual results; the MSE loss function is used, and the Adam optimizer is used. Record the training loss before enhancement for each Training. pre Enhanced test loss pre and enhanced training loss Train pos Enhanced test loss pos ;

[0017] Step 8: Calculate the training loss before the last e boosts (Train) pre Enhanced test loss pre Compared with augmented training loss pos Enhanced test loss posThe average value of each is obtained from ave(Train). pre ) e ,ave(Test pre ) e ,ave(Train pos ) e ,ave(Test pos ) e , e represents the set number of times, if the following conditions are met:

[0018] ave(Test pre ) e >ave(Test pos ) e

[0019] This indicates that the intelligent prediction method for battery health status based on random quantization data augmentation is effective and can better solve the SOH estimation problem when battery data is limited. If it does not meet the requirements, reset the number of segmentation intervals m in step 6 and repeat steps 7-9 until ave(Test) is satisfied. pre ) e >ave(Test pos ) e .

[0020] Step 4 specifically includes:

[0021] 4.1) Determine the number of intervals m for random quantization and the quantization intervals, obtain the minimum value min(x) and maximum value max(x) for each channel, and perform m uniformly distributed random samplings within the interval (min(x), max(x)) to obtain a. j And arrange them in ascending order to obtain a0, a1, a2, ... a m-1 Quantization interval S is obtained j =[a j a j+1 );

[0022] S j =[a j a j+1 )

[0023] a0, a1, a2, ...a m-1 =sort(a0′,a1′,a2′,...a m-1 ′)

[0024] a j ′=U(min(x),max(x)),j=0,1,2,...,m-1

[0025] 4.2) Map all values ​​within different quantization intervals to a single value yj, where y represents a uniformly distributed random sample within the corresponding interval:

[0026] y j =U(a j a j+1 )

[0027] 4.3) Scalar y obtained by quantization within different quantization intervals j Combine them and restore the processed data to the shape of the original data.

[0028] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0029] This invention discloses an intelligent prediction method for battery health status based on random quantization data augmentation. First, existing battery charging data is acquired through a Battery Management System (BMS), and the data is downsampled using linear interpolation to unify it to the same length, constructing original training and test sets. Then, a random quantizer is constructed to randomly quantize the original training set, generating a new training set, which is combined with the original training set to form an augmented training set. Finally, the augmented and unaugmented datasets are input into the constructed SOH prediction model, and the test losses for the augmented and unaugmented datasets are calculated respectively. If the augmented loss is less than the unaugmented loss, the intelligent prediction method based on random quantization data augmentation is considered effective. This invention effectively solves the accuracy problem of SOH estimation under limited battery data conditions.

[0030] This invention introduces a random quantization method to augment battery data, effectively alleviating the problem of low SOH prediction accuracy caused by limited battery data, thereby significantly improving the prediction performance of battery SOH. Attached Figure Description

[0031] Figure 1 This is a flowchart of a method according to an embodiment of the present invention. Detailed Implementation

[0032] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0033] Reference Figure 1 A smart prediction method for battery health status based on random quantization data augmentation includes the following steps:

[0034] Step 1: Obtain existing battery charging data via BMS, including voltage data (V). n，k，1 , ..., V n，k,i , ..., V n，k,l ), current data (I n，k，1 , ..., I n，k，i , ..., In，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;

[0035] Step 2: Perform linear interpolation downsampling on the existing battery charging data obtained in Step 1, downsampling all battery voltage data, current data, and time series data to the same data length l′;

[0036]

[0037] Among them, I p ′ represents the location of the sampling point, V n，k，p I n，k，p T n，k，p V represents the value of the p-th sampling point in the k-th cycle of the n-th battery after downsampling. n，k，i I n，k，i T n，k，i This represents the value of the i-th value of the n-th battery in the k-th cycle before downsampling, where i is I. p The value obtained by rounding down, I+1 is i p The value obtained by rounding up is...

[0038] Step 3: Construct the voltage, current, and time sample dataset O after linear interpolation downsampling. n ={O n，1 O n，2 O n，3 , ...O n，k , ...O n，K}, where O n，k O represents the k-th cycle of the n-th battery. n，k =(V n，k I n，k T n，k That is, each sample contains three channels, and the three channels store voltage data V respectively. n，k and current data I n，k and time data T n，k V n,k I n，k T n，k Each set consists of l′ data points; and the original training set Train and the test set Test are divided. One battery data point is selected as the test set Test, and the other battery data points are used for the original training set Train.

[0039] Step 4: Construct a random quantizer. The random quantizer dynamically quantizes the data in each channel, and the quantized values ​​are randomly sampled from randomly divided intervals. The specific steps are as follows:

[0040] 4.1) Determine the number of intervals m for random quantization and the quantization intervals, obtain the minimum value min(x) and maximum value max(x) for each channel, and perform m uniformly distributed random samplings within the interval (min(x), max(x)) to obtain a. j And arrange them in ascending order to obtain a0, a1, a2, ... a m-1 Quantization interval S is obtained j =[a j a j+1 );

[0041] S j =[a j a j+1 )

[0042] a0, a1, a2, ...a m-1 =sort(a0′,a1′,a2′,...a m-1 ′)

[0043] a j ′=U(min(x),max(x)),j=0,1,2,...,m-1

[0044] 4.2) Map all values ​​within different quantization intervals to a single value yj, where y represents a uniformly distributed random sample within the corresponding interval;

[0045] y j =U(a j a j+1 )

[0046] 4.3) Scalar y obtained by quantization within different quantization intervals j Combine them and restore the processed data to the shape of the original data;

[0047] Step 5: Determine the number of segmentation intervals m, and use a random quantizer to randomly quantize the original training set Train. The voltage, current, and time data in each sample are effectively preserved, generating enhanced samples and thus the generated training set Train′. Then, merge the original training set Train and the randomly quantized enhanced training set Train′ to obtain the enhanced training set Train. aug =Train∪Train′;

[0048] Step 6: Construct a battery state of health (SOH) prediction model. The battery state of health (SOH) prediction model is based on the ResNet network architecture, including five residual blocks and one predictor. Each residual block contains two convolutional layers, and batch normalization and ReLU activation functions are applied after the convolutional layers to ensure the training stability and non-linear expressive power of the model. Finally, the predictor with two fully connected layers maps the features extracted from the residual blocks to a scalar, which is used to output the battery state of health (SOH).

[0049] Step 7: Connect the original training set (Train) and the enhanced training set (Train) separately. aug =Train∪Train′ Input the battery state of health (SOH) prediction model into each of the following methods, obtain the SOH prediction results, and calculate the loss between the predicted and actual results; the MSE loss function is used, and the Adam optimizer is used. Record the training loss before enhancement for each Training. pre Enhanced test loss pre and enhanced training loss Train pos Enhanced test loss pos ;

[0050] Step 8: Calculate the training loss before the last e boosts (Train) pre Enhanced test loss pre Compared with augmented training loss pos Enhanced test loss pos The average value of each is obtained from ave(Train). pre ) e ,ave(Test pre ) e ,ave(Train pos ) e ,ave(Test pos ) e , e represents the set number of times, if the following conditions are met:

[0051] ave(Test pre ) e >ave(Test pos ) e

[0052] This indicates that the intelligent prediction method for battery health status based on random quantization data augmentation is effective and can better solve the SOH estimation problem when battery data is limited. If it does not meet the requirements, reset the number of segmentation intervals m in step 6 and repeat steps 7-9 until ave(Test) is satisfied. pre )e >ave(Test pos ) e .

[0053] The embodiments are based on a Tesla car battery dataset from a laboratory to verify the effectiveness of the method of the present invention. The dataset used for verification is a Tesla battery dataset from a laboratory. The nominal capacity of the battery is 2.7 Ah, the nominal voltage is 3.6 V, and the charging cut-off voltage and discharging cut-off voltage are 4.2 V and 2.0 V, respectively. The entire experiment was conducted at room temperature. Five batteries from a certain package were selected for k-fold cross-validation. One battery's data was selected as the test set, and the other battery data were used as the original training set. Five sets of experiments were repeated. The mean squared error (MSE) metric was used to measure the effect of the data augmentation of the present invention. The experimental results are shown in Table 1.

[0054] Table 1

[0055]

[0056] The results, converted to the form of SOH absolute error, are shown in Table 2:

[0057] Table 2

[0058]

[0059]

[0060] As can be seen from Tables 1 and 2, the test datasets generated by the datasets augmented using the random quantization method all have smaller losses than those before augmentation, indicating that the method proposed in this invention can effectively solve the SOH prediction problem under the condition of limited battery data.

[0061] This invention is applicable to SOH assessment when battery data is limited. It can enhance existing data through random quantization when battery data is limited, thereby improving the accuracy of model prediction of battery SOH to a certain extent.

Claims

1. A method for intelligent prediction of battery health status based on random quantization data augmentation, characterized in that, Includes the following steps: Step 1: Obtain existing battery charging data via BMS, including voltage data (V). n,k,1 ,…,V n,k,i ,…,V n,k,l ), current data (I n,k,1 ,…,I n,k,i ,…,I n,k,l ), and its corresponding time series is T. n,j,1 ,…,T n,k,i ,…,T n,k,l Where i∈1,…,l, represents the i-th data point in the k-th charging cycle of the n-th battery; Step 2: Perform linear interpolation downsampling on the existing battery charging data obtained in Step 1, downsampling all battery voltage data, current data, and time series data to the same data length l. ' ; Among them, i p 'Represents the location of the sampling point, V n,k,p ,I n,k,p ,T n,k,p V represents the value of the p-th sampling point in the k-th cycle of the n-th battery after downsampling. n,k,u ,I n,k,i ,T n,k,i This represents the value of the i-th value in the k-th cycle of the n-th battery before downsampling, where i is the value obtained by rounding down ip', and i+1 is the value of i. p The value obtained by rounding up is... Step 3: Construct the voltage, current, and time sample dataset O after linear interpolation downsampling. n ={O n,1 O n,2 O n,3 ,···O n,k ,···O n,K }, where O n,k O represents the k-th cycle of the n-th battery. n,k =(V n,k ,I n,k ,T n,k That is, each sample contains three channels, and the three channels store voltage data V respectively. n,k and current data I n,k and time data T n,k V n,k ,I n,k ,T n,k Each set consists of l' data points; and the original training set Train and the test set Test are divided, with one battery data point used as the test set Test and the other battery data used as the original training set Train. Step 4: Construct a random quantizer. The random quantizer dynamically quantizes the data in each channel, and the quantized values ​​are randomly sampled from randomly divided intervals; specifically: 4.1) Determine the number of intervals m for random quantization and the quantization intervals, obtain the minimum value min(x) and maximum value max(x) for each channel, and perform m uniformly distributed random samplings within the interval (min(x), max(x)) to obtain a. j And arrange them in ascending order to obtain a0, a1, a2, ... a m-1 Quantization interval S is obtained j =[a j ,a j+1 ); S j =[a j ,a j+1 ) a0, a1, a2,…a m-1 =sort(a0',a1',a2',…a m-1 ') a j '=U(min(x),max(x)),j=0,1,2,…,m-1 4.2) Map all values ​​within different quantization intervals to a single value y. j y j This represents a random sample that is uniformly distributed within the corresponding interval: y j =U(a j ,a j+1 ) 4.3) Scalar y obtained by quantization within different quantization intervals j Combine them and restore the processed data to the shape of the original data; Step 5: Determine the number of segmentation intervals m, and use a random quantizer to randomly quantize the original training set Train. The voltage, current, and time data in each sample are effectively preserved, generating enhanced samples and thus the generated training set Train'. Then, merge the original training set Train and the randomly quantized enhanced training set Train' to obtain the enhanced training set train. aug =Train∪Train'; Step 6: Construct a battery state of health (SOH) prediction model. The battery state of health (SOH) prediction model is based on the ResNet network architecture, including more than three residual blocks and a predictor. Each residual block contains two convolutional layers, and batch normalization and ReLU activation functions are applied after the convolutional layers. Finally, the predictor with two fully connected layers maps the features extracted from the residual blocks to a scalar, which is used to output the battery state of health (SOH). Step 7: Connect the original training set (Train) and the enhanced training set (Train) separately. aug =Train∪Train' Input the battery state of health (SOH) prediction model, obtain the SOH prediction results, and calculate the loss between the prediction results and the actual results; the loss function is the MSE loss function, and the optimizer is the Adam optimizer. Record the training loss before enhancement for each Training. pre Enhanced test loss pre and enhanced training loss Train pos Enhanced post-test loss Train pos ; Step 8: Calculate the training loss before the last e boosts (Train) pre Enhanced test loss pre Compared with augmented training loss pos Enhanced test loss pos The average value of each is obtained from ave(Train). pre ) e ,ave(Test pre ) e ,ave(Train pos ) e ,ave(Test pos ) e , e represents the set number of times, if the following conditions are met: ave(Test pre ) e >ave(Test pos ) e This indicates that the intelligent prediction method for battery health status based on random quantization data augmentation is effective and can better solve the SOH estimation problem when battery data is limited. If not, reset the number of segmentation intervals m in step 6 and repeat steps 7-9 until ave(Tesr) is satisfied. pre ) e >ave(Tesr pos ) e .

Images