基于适应度估计的动力电池冷却系统仿真优化加速方法

Simulation optimization and acceleration method of power batterycooling system based on fitness estimation

动力电池风冷系统仿真设计优化过程中存在“软件仿真+进化算法”的技术路线效率较低和难以被接受的问题,针对该问题提出一种基于适应度估计的动力电池冷却系统仿真优化加速方法。采用AP聚类将种群中出现的设计方案(个体)划分为多个子种群,通过软件仿真对子种群中代表性个体的适应度进行仿真计算,采取适应度估计来获取种群中非代表性个体的适应度。实验结果表明,该算法能有效地优化动力电池风冷系统的设计参数,大大缩短系统的优化时间。当目标函数设定为最高温度、平均温度时,比传统的“Fluent仿真+遗传算法”技术路线在取得相似优化结果的前提下,运行时间可减少70.5%。

In the process of simulation design optimization of power battery air cooling system,the technical route of“software simulation+evolution algorithm”is not efficient and thus has a low popularity.This paper proposes a method of simulation optimization and acceleration of power battery cooling system based on fitness estimation.In this method,the design schemes(individuals)in the population are divided into multiple subpopulations by AP clustering,and the fitness of representative individuals in the subpopulations is simulated by software simulation,and the fitness of non-representative individuals in the population is obtained by fitness estimation.Our experimental results show the algorithm effectively optimizes the design parameters of the power battery air cooling system and greatly cuts the optimization time of the system.When the objective function is set to the highest and average temperatures,the running time is reduced by 70.5%compared with the traditional“Fluent simulation+genetic algorithm”technical route under the premise of achieving similar optimization results.