化工学报 ›› 2021, Vol. 72 ›› Issue (3): 1512-1520.DOI: 10.11949/0438-1157.20201804

• 过程系统工程 • 上一篇    下一篇

基于改进差分进化算法的质子交换膜燃料电池模型参数优化识别

徐斌()   

  1. 上海工程技术大学机械与汽车工程学院,上海 201620
  • 收稿日期:2020-12-07 修回日期:2020-12-15 出版日期:2021-03-05 发布日期:2021-03-05
  • 通讯作者: 徐斌
  • 基金资助:
    国家自然科学基金项目(61703268)

Parameter optimal identification of proton exchange membrane fuel cell model based on an improved differential evolution algorithm

XU Bin()   

  1. School of Mechanical and Automotive Engineering, Shanghai University of Engineering Science, Shanghai 201620, China
  • Received:2020-12-07 Revised:2020-12-15 Online:2021-03-05 Published:2021-03-05
  • Contact: XU Bin

摘要:

质子交换膜燃料电池(proton exchange membrane fuel cell, PEMFC)模型中通常包含一些未知的难以确定的参数。为了有效确定这些参数,提出一种基于概率选择模型的改进差分进化算法。该概率选择模型为进化群体中的每个个体分配关于个体优劣性能的选择概率,然后基于该选择概率选择优秀的个体参与变异和交叉操作。为了验证改进算法的有效性,首先将其用于求解标准测试函数,实验结果表明,基于概率选择模型的变异、交叉操作能显著提高差分进化算法的收敛速度和求解精度。然后将改进算法用于质子交换膜燃料电池模型最优参数识别问题,得到的仿真结果和实验测试数据之间具有较高的拟合精度,表明本文改进算法是一种有效的求解系统模型参数优化识别的方法。

关键词: 燃料电池, 参数识别, 优化, 差分进化, 概率模型

Abstract:

The proton exchange membrane fuel cell (PEMFC) model usually contains some unknown parameters that are difficult to determine. To extract the parameters of PEMFC model exactly and accurately, in this paper, an improved differential evolution (IDE) is proposed based on probability-selection model. During the evolution process, the probability-selection model assigns a selection probability value that related to the fitness values for each solution in the main population. Based on these probability values, some superior solutions are selected as parent solutions during the mutation and crossover stages. Experiments over some complex benchmark test functions indicate that the proposed IDE method performs better than original differential evolution in terms of convergence speed and accusation indictors. When the proposed IDE is applied to solve parameter optimal identification problem of PEMFC system, experimental results show that the obtained fitting accuracy is acceptable, which means IDE is an efficient method to identify the parameters of PEMFC model.

Key words: fuel cells, parameter identification, optimization, differential evolution, probability-based model

中图分类号: