基于网络学习控制的燃料电池测试系统研究及应用

氢-氧燃料电池测试系统对温度、压力、流量、湿度等系统控制参数的控制精度、稳定性和鲁棒性的要求很高,而温度、湿度、流量等参数具有滞后、非线性且相互耦合的特性,因而单独对其进行控制是很难的.为了提高燃料电池的性能,提出了一种基于网络学习的控制策略:对温度、流量基本的控制由其自身的控制器完成,系统参数的优化,协调控制由网络学习单元完成,同时网络学习单元根据现场运行的实际情况实时优化和自适应调节本地控制器参数.同时,在网络学习单元建立了一个网络服务器,能够让用户在远程进行燃料电池测试试验,并可浏览相关的信息,真正实现了"管控一体化"和打破了"自动化孤岛"的现象.此外,对燃料电池进行了阻抗分析,为燃料电池的设计提供了依据.现场运行表明,本测试系统具有良好的稳定性和鲁棒性,能够对燃料电池进行有效的测试及分析.

Research and Application of Fuel Cell Stack Test System Based on Networked Learning Control

In fuel cell test system, the control precision, stability and robust of system control parameters including temperature, pressure, flow and humidity are required highly. Since the parameters are delay, nonlinear and coupling, separate control is very difficult. In order to improve fuel cell ability, networked learning control strategy is put up. The basic control of temperature and flow is accomplished by themselves' controller, the optimization of system parameters and harmonic control are performed by networked learning unit. At the same time, networked learning unit real-time optimizes and self-adaptively adjusts local controller parameters depend on field run-time condition. The user can remotely do fuel cell test experiments and browse correlative information by setting up a network server in networked learning unit. The control strategy realizes a complete unit of management and control, and breaks up "automation isolation". Furthermore, the impedance analysis supplies the decision of fuel cell. Field run shows that this test system has good stability and robust, and can effectively test and analyze fuel cell.