State-of-Charge Estimation of Li-ion Batteries Based on A Hybrid Model Using Nonlinear Autoregressive Exogenous Neural Networks
- Resource Type
- Conference
- Authors
- Zhang, Yuxiang; Zhao, Chunyu; Zhu, Senlin
- Source
- 2018 IEEE PES Asia-Pacific Power and Energy Engineering Conference (APPEEC) Power and Energy Engineering Conference (APPEEC), 2018 IEEE PES Asia-Pacific. :772-777 Oct, 2018
- Subject
- Power, Energy and Industry Applications
State of charge
Neural networks
Estimation
Training
Electronic countermeasures
Lithium-ion batteries
lithium-ion battery
battery management system (BMS)
state-of-charge (SOC) estimation
nonlinear autoregressive exogenous (NARX) neural network
extended Kalman filter (EKF)
- Language
Lithium-ion batteries are currently the most prevalent energy storage devices. Accurate estimation of battery state-of-charge (SOC) is an important task in the battery management system for ensuring safety operations of battery packs. This paper presents a hybrid model which substitutes a nonlinear autoregressive exogenous (NARX) artificial neural network for traditional RC networks to represent the polarization effect of lithium-ion batteries. Based on the hybrid model, the extended Kalman filter method is adopted to estimate the battery SOC. Experiments on a lithium-ion battery cell were conducted to verify the feasibility and effectiveness of this method. The results show that the hybrid model has a good performance in modeling battery dynamic characteristics, and whereby the estimation of battery SOC achieves high accuracy and convergence speed.