Improved PSO-TCN model for SOH estimation based on accelerated aging test for large capacity energy storage batteries

TY - JOUR

T1 - Improved PSO-TCN model for SOH estimation based on accelerated aging test for large capacity energy storage batteries

AU - Yu, Peiwen

AU - Zhou, Chidong

AU - Yu, Yajuan

AU - Chang, Zeyu

AU - Li, Xi

AU - Huang, Kai

AU - Yu, Juan

AU - Yan, Kang

AU - Jiang, Xiaoping

AU - Su, Yuefeng

N1 - Publisher Copyright: © 2024 Elsevier Ltd

PY - 2025/2/1

Y1 - 2025/2/1

N2 - The accurate estimation of the state of health (SOH) of lithium-ion batteries is crucial for enhancing the reliability and safety of battery systems. However, the current SOH estimation methods for large capacity lithium-ion energy storage batteries still face problems such as unsatisfactory estimation accuracy. Therefore, this paper proposes a method for estimating the state of health through multi-health features extraction combining temporal convolutional network and particle swarm optimization. In order to accurately describe the accelerated aging mechanism of large capacity energy storage batteries, various health features are extracted from battery data, such as time features, energy features, capacity features, and incremental capacity features. The grey correlation analysis method is used to evaluate the correlation between health features and SOH. In order to overcome the difficulty of selecting hyper-parameters for neural network models, a particle swarm optimization algorithm and a learning rate scheduler are proposed to correctly obtain hyper-parameters and achieve accurate estimation of battery SOH. In order to overcome the difficulty of selecting hyper-parameters for neural network models and dynamically adjust the learning rate to meet the learning needs of the model at different training stages, a particle swarm optimization algorithm and learning rate scheduler are proposed to correctly obtain hyper-parameters and achieve accurate estimation of battery SOH. The experimental results show that the mean absolute error and root mean square error of this method are both within 2 %, and it has high SOH accuracy and robustness.

AB - The accurate estimation of the state of health (SOH) of lithium-ion batteries is crucial for enhancing the reliability and safety of battery systems. However, the current SOH estimation methods for large capacity lithium-ion energy storage batteries still face problems such as unsatisfactory estimation accuracy. Therefore, this paper proposes a method for estimating the state of health through multi-health features extraction combining temporal convolutional network and particle swarm optimization. In order to accurately describe the accelerated aging mechanism of large capacity energy storage batteries, various health features are extracted from battery data, such as time features, energy features, capacity features, and incremental capacity features. The grey correlation analysis method is used to evaluate the correlation between health features and SOH. In order to overcome the difficulty of selecting hyper-parameters for neural network models, a particle swarm optimization algorithm and a learning rate scheduler are proposed to correctly obtain hyper-parameters and achieve accurate estimation of battery SOH. In order to overcome the difficulty of selecting hyper-parameters for neural network models and dynamically adjust the learning rate to meet the learning needs of the model at different training stages, a particle swarm optimization algorithm and learning rate scheduler are proposed to correctly obtain hyper-parameters and achieve accurate estimation of battery SOH. The experimental results show that the mean absolute error and root mean square error of this method are both within 2 %, and it has high SOH accuracy and robustness.

KW - Accelerated aging test

KW - Large capacity energy storage batteries

KW - PSO

KW - State of health

KW - TCN

KW - lithium-ion battery

UR - https://www.scopus.com/pages/publications/85212340985

U2 - 10.1016/j.est.2024.115031

DO - 10.1016/j.est.2024.115031

M3 - Article

AN - SCOPUS:85212340985

SN - 2352-152X

VL - 108

JO - Journal of Energy Storage

JF - Journal of Energy Storage

M1 - 115031

ER -