基于启发式动态规划的固体氧化物燃料电池优化控制研究  

Study on Optimization Control of Solid Oxide Fuel Cell Via Heuristic Dynamic Programming

作　　者：宋春宁[1] 钟文瀚[1] 

机构地区：[1]广西大学电气工程学院,南宁530004

出　　处：《计算机测量与控制》2012年第7期1830-1833,共4页Computer Measurement &Control

基　　金：国家自然科学基金项目(60964002);广西教育厅科研项目(200911MS11)

摘　　要：固体氧化物燃料电池(SOFC)系统是一个非线性系统,现存的建模方法和优化控制算法很难对其进行精确的建模及优化控制;针对此问题,采用基于数据的建模方法,对固体氧化物燃料电池系统进行BP神经网络建模,然后在此基础上,首次采用启发式动态规划(HDP)算法对固体氧化物燃料电池系统中的各种气体分压、输出电压以及温度进行优化控制;Matlab仿真结果表明,基于BP神经网络的HDP优化算法具有收敛速度快、鲁棒性强、控制精度高等优点,并使固体氧化物燃料电池系统在负载变化时很快稳定输出电压,实现了优化控制,减少能耗。Abstract.. Solid oxide fuel cell （SOFC） system is a nonlinear system. It is hard to build accurate mathematical model and achieve optimal control for it. To This Question, the idea is building a BP neural network model of solid oxide fuel cell system based on the data of the model- ing method, then based on this, performing firstly the optimal control to the solid oxide fuel cell system of all kinds of gas pressure, output voltage and temperature by the heuristic dynamic programming （HDP） algorithm. The result of Matlab simulation shows that the optimiza- tion algorithm of the HDP based on the BP neural network implement the optimal control of the solid oxide fuel cell system with strong ro- bustness, fast convergence Speed and high control precision, and make solid oxide fuel cell system quickly stable output voltage when the load is changing, realize the optimal control, reduce energy consumption.

关 键 词：固体氧化物燃料电池(SOFC) BP神经网络 HDP算法 优化控制 

分 类 号：TP273.2[自动化与计算机技术—检测技术与自动化装置]