繁殖型PSO算法在电池模型辨识中的应用

Application of PSO algorithm with evolutionary mechanism in battery parameter identification

针对铅酸蓄电池在工作中呈现非线性特性,电池模型等效参数随其荷电状态(SOC)改变而发生变化以致难以准确估计的问题,本文结合常见的等效电池模型,应用一种带有繁殖机制的粒子群优化(PSO)算法对蓄电池模型参数进行了辨识。本研究采用基于特定分析方程的参数描述形式,在不同SOC状态下对等效模型进行了参数优化。测试结果表明,采用这种带有繁殖机制的PSO算法所估计出的蓄电池模型能够较准确地跟踪电池的实际工作电压,从而验证了该算法在蓄电池模型辨识中的实用价值,为建立准确的蓄电池模型提供了一个系统化、理论化的方法。

In practical applications, battery usually works in a nonlinear state and its equivalent model parameters change with battery state of charge （ SOC） . In order to correctly estimate the model parameter of valve regulated lead-acid battery, a particle swarm optimization （ PSO） with evolutionary mechanism is employed based on a com-monly used battery model. In this paper a specific empirical function is adopted to describe the parameter-SOC re-lation, and the parameters under different SOC states are identified. Identification results are compared and it is shown that with this hybrid PSO algorithm, the identified battery model can track the practical working performance reasonably well, so a strong practicability of this algorithm is verified, and a systematic method for battery parame-ter identification is established.