摘要
为了消除神经网络参数初值对控制器性能的影响,提出了一种改进遗传算法优化的RBF神经网络控制器。该方法设计了基于性能指标的适应度函数,自适应的交叉概率、变异概率,引入移民的遗传算法,保证了得到的控制器为最优参数控制器。该方法可用于非线性对象的控制器设计,仿真结果说明了该方法的有效性。
To eliminate the influence of initial weights values of neural networks on controllers, a RBF neural network controller optimized by a new improved genetic algorithm （GA） is proposed. In the improved genetic algorithm, it designed special fitness function based on performance of controller, and adaptive probability of crossover and mutation, and improved immigration method. The improved GA ensures that the controller with the parameters based on the GA is optimal. The proposed method can be used to design controllers for nonlinear plants. The simulation results show the effectiveness of proposed controller.
出处
《电机与控制学报》
EI
CSCD
北大核心
2007年第2期183-187,共5页
Electric Machines and Control
基金
国家自然科学基金项目(60325311
60534010
60572070)
国家教委博士点基金项目(2001045023)
沈阳市自然科学基金项目(1022033-1-07)
关键词
RBF神经网络
遗传算法
移民方法
RBF neural network
genetic algorithm
immigration method
