高斯基函数CMAC快速算法的改进及应用研究

Improvement of Speedy Algorithm of CMAC for Gauss Basis Functions and Its Application

该文对基于高斯基函数小脑模型(CMAC)的快速算法进行了改进,针对其学习速率的选取问题,提出了一种基于遗传算法的学习速率最优选取方法,使得CMAC学习速率的选取得到了最优化。讨论了该算法的实际可行性,提出了参数选择和实时控制相分离的策略,并在某转台伺服系统模型中进行了应用研究。仿真结果表明,改进算法避免了学习速率选取的经验不确定性。

The speedy algorithm of Gauss Basis CMAC(Cerebellar Model Articulation Controleer) is improved.Inview of its problem of learning rate's selection,an optimization of learning rate based on Genetic Algorithm is presented.This method makes the learning rate's selection be optimal.The feasibility of this method is discussed.A tactic of separating the process of the parameter optimization from the practical control process is presented.The simulation results in a servo system show that this improved method can avoid the uncertainty of learning rate selected by experience and improve the speed of convergence of the CMAC.