Article, 2024

Adaptive Kalman filter and self-designed early stopping strategy optimized convolutional neural network for state of energy estimation of lithium-ion battery in complex temperature environment

Journal of Energy Storage, ISSN 2352-152X, 2352-1538, Volume 83, Page 110750, 10.1016/j.est.2024.110750

Contributors

Li, Jin [1] Wang, Shunli 0000-0003-0485-8082 (Corresponding author) [1] [2] Chen, Lei [1] Wang, Yangtao [1] Zhou, Heng [1] Guerrero, Josep M 0000-0001-5236-4592 [3]

Affiliations

  1. [1] Southwest University of Science and Technology
    2. [NORA names: China; Asia, East];
  2. [2] Inner Mongolia University of Technology
    3. [NORA names: China; Asia, East];
  3. [3] Aalborg University
    4. [NORA names: AAU Aalborg University; University; Denmark; Europe, EU; Nordic; OECD]

Abstract

To achieve accurate State of Energy (SOE) estimation of Battery Management System (BMS), the Adaptive Kalman Filter and self-designed Early Stopping Optimized Convolutional Neural Network (AKF-ESCNN) is innovatively introduced. It is based on a self-designed Early Stopping (ES) strategy to optimize the training of Convolutional Neural Network (CNN) models, addressing the issue of network overfitting. By integrating Adaptive Kalman Filtering (AKF) for smoothing and filtering the network outputs, it reduces erroneous abrupt variations in results, ultimately achieving precise estimation of SOE. After different experimental data verification (5 °C, 10 °C and 25 °C), compared the loss values of model training. AKF-ESCNN model training accuracy is 10 % higher than CNN. In the whole temperature range of this paper, AKF-ESCNN also has a better performance. At cold −5 °C the Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE) of AKF-ESCNN in the HPPC working condition are 0.268 % and 0.449 %, while the MAE and RMSE of CNN before optimization are 1.411 % and 1.973 %, and the estimation accuracy has been significantly improved. AKF-ESCNN provides a new way to solve the problems faced by data-driven SOE estimation of lithium-ion batteries.

Keywords

HPPC, Kalman filter, SOEs, abrupt variations, absolute error, accuracy, accurate state, adaptive Kalman filter, battery, battery management system, complex temperature environment, conditions, convolutional neural network, data verification, early stopping, early stopping strategy, environment, error, estimation, estimation accuracy, estimation of lithium-ion batteries, experimental data verification, filter, issues, lithium-ion batteries, loss, loss values, management system, mean, mean absolute error, mean square error, model, model training, model training accuracy, network, network output, network overfitting, neural network, optimization, optimized Convolutional Neural Network, output, overfitting, performance, problem, range, results, root, root mean square error, smoothing, state, state of energy (SOE, stop, stopping strategy, strategies, system, temperature environment, temperature range, training, training accuracy, training of convolutional neural networks, variation, verification, whole temperature range, working conditions

Funders

  • National Natural Science Foundation of China

Data Provider: Digital Science

LINKS
-

Matching Records in NORA

SUBJECTS
+

METRICS
+