Application of Gaussian-sequential-probabilistic-inference concept based Kalman filter for accurate estimation of state of charge of battery

摘要

Advanced technology and research for lithium-ion (Li-ion) battery charging and discharging within operating range of State of charge (SOC) is a great challenge. Due to the complex operational environment and noise, it is difficult to estimate Li-ion battery voltage, current and terminal voltage accurately, which are required to SOC estimation. To address these issues, Gaussian-Sequential-Probabilistic-Inference Concept based Kalman filter (GSPIC-KF) for Li-ion SOC estimation is proposed in this paper. This paper presents a detail design and analysis of GSPIC-KF, which is used to find a state estimate and minimize error between the true and estimated states. Proposed technique uses a recursion to compute new estimate based on the prior estimate. It results, sequence of estimate provided by a sequence of steps by considering the randomness of the process noise and the sensor noise. The improved performance of the developed system with GSPIC-KF is verified through a detailed comparative study among Coulomb counting (CC), extended Kalman filter (EKF), artificial neural network (ANN), Voltage based method (VBM) and GSPIC-KF. A comparison study results better performance of GSPIC-KF method over others in terms of convergence time, accuracy, boundary condition, error, precision, and noise rejection. This paper also represents Li-ion battery model design and analysis with proper estimation of SOC to obtain optimal performance and prolong life time. The Peukert's method is used here for battery modelling and battery capacity estimation. An experimental prototype is developed in the laboratory to monitor the perfor-mance of the individual cells and 48 V, 40 Ah Li-ion battery pack. The performance of battery pack is tested experimentally under different operating conditions. Real time observation and an accurate setting of parameters for the proper operation of battery and battery management system (BMS) are presented in the paper.