用过程神经网络和遗传算法实现系统逆向求解

Realization of system converse solution based on process neural networks and genetic algorithm

中文关键词  过程神经网络  遗传算法  过程控制  逆向求解

英文关键词  process neural networks  genetic algorithm  process control  converse solution

中文摘要

      对于多输入多输出系统,针对如何根据系统模型和期望输出反求系统输入的问题,本文提出了一种基于过程神经网络和遗传算法相结合的方法.首先根据实际系统的领域知识和学习样本集,建立满足系统实际输入输出映射关系的正向过程神经网络.然后按照系统在过程区间的某一期望输出,用过程神经网络的输出误差构造适应度函数,用遗传算法逆向确定系统的过程输入信号,使该输入信号满足已建立的正向过程映射关系,从而完成系统的逆向过程控制.文中给出了具体的实现算法并给出了此方法的一个应用实例.

英文摘要

      A optimization algorithm of process neural networks and genetic algorithm(PNN-GA) is proposed,it aims at determining MIMO system input by the system model and desired output.First,the process neural networks(PNN) that represent the mapping relation between input and output of the system is founded according to system field knowledge and training sample sets.Secondly,fitness function of genetic algorithm(GA) is constructed according to PNN output error.Based on the desired output of the system,we determined the process input signal which conforms to the PNN mapping relation that was found,thus the system converse process solution is accomplished.The general realization approach is presented in this paper.An application example is given to illustrate the applicability of the approach.