基于RLSSVM-CPSOSA的PEM燃料电池系统建模  

Modeling of PEMFC stack based on RLSSVM-CPSOSA algorithm

作　　者：赵超[1] 陈培江 王贵评 戴坤成[1] 

机构地区：[1]福州大学石油化工学院,福建福州350108

出　　处：《计算机与应用化学》2017年第3期201-206,共6页Computers and Applied Chemistry

基　　金：国家自然科学基金项目(60804027;61374133);高校博士点专项科研基金(20133314120004)

摘　　要：PEMFC电堆是一个包括流动、传热、传质和电化学反应等多种物理化学现象的复杂机体,其建模问题是一个具有挑战性的问题。本文提出一种基于RLSSVM-CPSOSA模型的燃料电池建模方法。CPSOSA算法将CO算法、模拟退火算法和粒子群算法有机结合,以克服粒子群算法早熟收敛的不足。然后利用CPSOSA算法对鲁棒最小二乘支持向量机模型(RLSSVM)进行参数寻优,从而获得RLSSVM-CPSOSA燃料电池模型。以MATLAB平台搭建PEMFC电堆模型并进行仿真研究,结果表明所提出的RLSSVM-CPSOSA模型的有效性和良好的预测精度。为PEM燃料电池实时控制系统奠定了基础。A PEMFC stack is a complex object consisted of a lot of physicochemical characteristics such as fluxion, heat transfer, mass transfer process and electrochemical reaction, and modeling of PEMFC stack is a difficult and unsolved subject at present. A RLSSVM-CPSOSA method based on robust least squares support vector machine and CPSOSA algorithm is developed for modeling of PEMFC stack. By integrating chaos optimization （CO） and simulated annealing （SA） to the PSO, the CPSOSA algorithm, it makes the particles of CPSOSA maintain the diversity during the evolution so as to overcome the premature convergence of PSO. Then, the proposed CPSOSA algorithm is applied to obtain the optimal parameters of the RLSSVM. Researches on the optimized model are illustrated based on the data from a simulated PEMFC stack, and the results show that the proposed approach has great estimation accuracy and validity.

关 键 词：质子交换膜燃料电池 鲁棒最小二乘支持向量机 混沌粒子群模拟退火算法 

分 类 号：TQ015.9[化学工程] TP391.9[自动化与计算机技术—计算机应用技术]