Model-based and data-driven state of charge estimation of Li-ion battery

• Main Author: Chen, Andi
• Other Authors: Xu Yan
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Engineering; Li-ion battery; State of charge; Battery model; Data-driven
• Online Access: https://hdl.handle.net/10356/181141
• Institution: Nanyang Technological University

Accurate state of charge (SOC) estimation is essential for optimizing the performance and improving the reliability of battery energy storage systems (BESS). This dissertation focuses on comparing three distinct algorithms: Extended Kalman Filter (EKF), Particle Filter (PF), and Long Short-Term Memory (LSTM) networks. by utilizing a second-order Thevenin equivalent circuit model (ECM), we implement parameter identification through the Particle Swarm Optimization (PSO) algorithm, which demonstrates efficiency and accuracy comparable to traditional methods. The HPPC data used for parameter identification and simulated EV driving data under four operating conditions of DST, FUDS, NEDC, and WLTP are obtained from the battery test platform of the Stability, Optimization, and Data-Analytics (SODA) power system research group, Nanyang Technological University. The outcomes show that the PF algorithm performs noticeably better than the other two, achieving the average error of 0.240% and the maximum error of 0.610%, while the EKF and LSTM exhibit higher errors. The experimental findings confirm that the PF algorithm not only provides greater accuracy but also maintains stability across different scenarios, making it a robust choice for practical applications. This dissertation highlights the benefits of the PF algorithm while adding to our understanding of SOC estimating techniques. The comprehensive analysis of the algorithms, combined with the rigorous modeling of Li-ion batteries, establishes a framework for future research and development in battery technology, particularly focusing on enhancing estimation techniques for SOC and ensuring optimal performance of battery systems in real-world applications.