Study On Models And Methods For Online Estimation Of State Of Charge Of Power Battery Pack

With the rapid development of the national economy and increasing living standards in recent years, the automotive industry has been developing rapidly in the world. A large number of cars bring convenience to people’s lives, but also has brought many negative effects of energy consumption, and environmental pollution. With these growing excessive oil consumption and environmental issues, Governments in the world coincidently develop electric vehicles (EVs) as a new green transport. In electric vehicles, batteries are used directly as the active energy supply, their working states are very important to driving safety and operational reliability of the whole car. To ensure good performance of the battery pack in an EV so as to extend its service life, a timely and accurate understanding of the various operating status, especially the state of charge (SOC) of the battery pack, are very critical to the whole system. In this paper, the problem of online accurate estimate of the SOC for a lithium power battery pack is considered. Extensive researches on the dynamic lithium battery models and the algorithms for SOC estimation are carried out. The main research works of the thesis are as follows:(1)For the power battery dynamic nonlinear system, specific improvements of the process model of lithium batteries are proposed, and the corresponding model parameters estimation algorithms are given. A discharge rate proportional coefficient is used to model the relationship between different discharge rates and the SOC of the battery, which is a second order polynomial; on the other hand, a temperature proportional coefficient is utilized to model the relationship between different temperature conditions and the SOC, which is another second-order polynomial. The combination of the discharge rate proportional coefficient and the temperature proportional coefficient can effectively simulate the discharge characteristics of lithium batteries in real operating conditions, and can effectively improve the estimated accuracy of the SOC of a single battery.(2)Bayesian filtering methods such as the extended Kalman filter, the Unscented Kalman filter and the particle filter, are used for battery SOC estimation. Algorithms and specific steps are given in detail. Several typical operating conditions are simulated to verify the effectiveness of the proposed methods. Comparative analysis for the SOC estimation performances of the SOC estimatin algorithms such as the accuracy, convergence speed, complexity and robustness are also shown.(3)Individual differences exist inevitablely during production processe or using process. On the other hand, aging effect exist commonly for batteries. All these make the battery model parameters variate timely. In this paper, an online joint estimation method of the lithium battery SOC and the battery model parameters, especially the internal resistance, are proposed, which can further improve the estimation accuracy of the battery SOC.(4)A prototype demonstration system of SOC online estimation for battery pack is developed. The battery pack is composed of four lithium-ion batteries with a nominal capacity of50Ampere hours and a nominal voltage of3.2volts. Both the system hardware and software are built. Experimental results show that the sigma point Kalman filter based internal resistance and SOC joint estimation algorithm can be used for a precise estimation of the SOC of the battery pack with a maximum estimation error of5%, while the average estimation error is3%. The time consumed for one step estimation is3～4seconds. The developed embedded prototype system verifies that the proposed methods can be applied for real applications.