一种基于改进遗传算法的模糊神经网络控制器及其在烧结终点控制中的应用

针对烧结过程这一复杂、多参数耦合的高度非线性系统，融合遗传算法、神经网络和模糊控制的优点，提出一种基于改进遗传算法的模糊神经网络控制方法，并应用于烧结过程终点控制．首先采用遗传算法对给定的模糊神经网络控制器结构参数进行离线优化，然后利用BP算法较强的局部搜索能力和对对象的适应能力，进一步进行参数的在线调整．同时，为解决传统遗传算法早熟和收敛速度慢的问题，从交叉和变异算子、适应度函数选取等方面对遗传算法进行改进．采用精英保留策略，提高了全局搜索性能和收敛速度．仿真结果表明，所提出的控制器优于常规的模糊神经网络控制器(Fuzzy Neural Network Controller, FNNC)．算法的实际应用效果良好，为解决烧结终点控制问题提供了一条新的途径．

An Improved Genetic Algorithm Based Fuzzy Neural Network Controller and Its Application to Burning Through Point Control

By integrating the advantages of genetic algorithm (GA),neural network and fuzzy control,a fuzzy neural network controller based on improved GA is proposed for sintering process which is a highly nonlinear system with complexity and multiple parameters,and the controller is used to control the burning through point(BTP) in sintering process.Structural parameters of the fuzzy neural network are firstly optimized off-line by GA,and then are adjusted on-line by using the local searching and adaptabilities of B...