基于贝叶斯次优解的锂电池SOC初值追踪研究

在使用循环算法的剩余电量(state of charge,SOC)估算的理论研究中,通常都是设定已知的初始值,无法完全论证算法的各项性能指标.在研究剩余电量估算问题时,则重点针对不确定的初始赋值,使用一种贝叶斯次优解法——粒子滤波(PF),基于SOC动态观测模型,研究其收敛到真值的情况,并辅以MATLAB仿真实验.结果显示在非平台期PF可快速追踪初值,而在平台区由于自身算法的缺陷,无法追踪初值,应根据此区域的电压特性,更换为另一种次优解算法.论证了在解决锂电池初值追踪问题上,两种次优解算法的结合使用可以得到理想结果.

Lithium-ion Batteries SOC Initial Value Tracking Based on Suboptimal Solutions of Bayesian Filter

In the theoretical study of the loop algorithm for estimating state of charge(SOC),it is usually set the initial value that is known and cannot fully demonstrate the performance indexes of the algorithm.While this paper aims at the problem about uncertain initial assignment and studies its convergence to the true value based on dynamic observation model using particle filter,supplemented by MATLAB simulation.The results show that satisfactory results are obtained only in the area where is out plateau.We can not track the initial value in plateau because of the flaw of algorithm.Another suboptimal solution should be used based on the voltage characteristic in this area.It demonstrates that two suboptimal solutions should be combined to obtain the ideal results.