神经网络非线性逆控制方法实现PEMFC分布式发电系统的负荷响应

质子交换膜燃料电池(PEMFC)分布式发电系统的负载典型特点是时时有瞬间的大的峰值功率。因此,系统必须找到实时的最佳工作点才能响应负载的功率需求。以往采用的优化搜索方法需要在负载连续变化和峰值现象之间权衡搜索步长,影响控制的速度和精度。分析PEMFC分布式发电系统的工作特点以及各工作参数与净输出功率之间的关系,提出利用BP神经网络实现PEMFC输出功率的复杂非线性逆模型。在此基础上,根据非线性逆控制的思想设计了系统实时功率响应的逆控制器。经分析和仿真验证,利用神经网络非线性逆控制方法,PEMFC系统在稳定运行过程中能够满足系统内部功率损耗的同时良好的响应系统负载的实时功率需求。与优化搜索方法比较,该实时控制设计在实现精度和速度上都有所改进。图5表1参6

Neural- Network- Nonlinear-Inverse- Based Control Strategy for Realizing Real-Time Load Responses in Distributed PEMFC Power Systems

It is typical for PEMFC distributed power generating systems that they may incur instantaneous load demands at any time. It is therefore imperative for such systems to detect the optimal real time working point, enabling them to respond to load demands. The optimizing searching method previously used, because of its having to find a suitable step length that meets both the requirements of continous load and peak demands, affects the speed and precision of real time control. Following an analysis of the characteristics of PEMFC distributed power generating systems and of the relationship between operation parameters and net power output, a proposal is presented for constituting a complex nonlinear inverse based model of PEMFC's power output by making use of BP neural networks. Herewith, and according to the concept of nonlinear inverse control, an inverse controller of the system's real time response has been designed. Analysis and probation by simulation show that, during steady operation, nonlinear inverse control, based on neural networks, can make PEMFC systems readily respond to the system's real time load requirements while simultaneously making up for the system's internal power consumption. Compared with the optimizing searching method, this real time control design exhibits improvements both in respect to precision and to speed. Figs 5, table 1 and refs 6.