摘要
广义小脑模型（ＣＭＡＣ）在函数泛化能力和函数逼近能力方面优于基本ＣＭＡＣ模型，但算法较为复杂，实时性差．因此，研究广义ＣＭＡＣ模型的快速算法，对于满足实时控制是非常必要的．文中研究了基于高斯基函数的广义ＣＭＡＣ模型的快速算法，定义了包含待学习样本点的一个超立方体子空间，提出了基于该超立方体子空间的快速学习算法．通过算例仿真表明，学习算法收敛速度较快，可以满足实时控制要求．
The general CMAC model has advantages over the conventional CMAC model both in function generalization and function approximation capability, but the algorithms are more complicated and can not be used in realtime control system. Therefore, it is necessary to study speedy algorithms for the general CMAC to meet the requirements of realtime control. In this paper, the speedy algorithms for the general CMAC with Gaussian basis functions are investigated. A hypercube subspace which covers the samples to be learned is defined, and a novel approach of speedy learning algorithm is proposed based on the hypercube subspace. The simulation results show that this algorithm has a high convergence speed and satisfies the requirements of realtime control.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
1998年第8期63-65,共3页
Journal of Shanghai Jiaotong University
关键词
高斯基函数
广义小脑模型
实时控制系统
算法
Gauss basis function
general cerebella model articulation controller (CMAC)
hypercube subspace
