摘要
在ＣＭＡＣ算法的基础上，采用两种方法来克服其存储容量随分辨力剧增的缺陷，一种方法是针对输入数据集分布和量化级一致的情况下，利用变分法，求得最佳非均匀量化曲线，使得量化噪声均方值最小；另一种是在输入数据集分布未知，量化级给定的情况下，利用求重心的方法，提高网络分辨力，从而避免了以增大存储容量来提高分辨力，大大提高了网络的分辨力与推广能力，使该算法更为实用．采用第二种方法，应用于四足步行机器人伺服系统的模糊控制。
CMAC(cerebellar model articulation controller) neural network requires much more memory when the resolving power enhances. If the distribution of the input data set and the level of the qualification are given, the optimum qualification method based on calculus of variations can be used to minimize the mean square error. If the distribution of the input data set is unknown and the level of the qualification is given, it uses the method for gravity center calculating. These methods can increase the resolving and generalization ability of the network and make the network more practicable. The latter method can be applied to the fuzzy control of the quadruped robot's servo system and has a good effect.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
1999年第10期1276-1279,共4页
Journal of Shanghai Jiaotong University
基金
国家自然科学基金
关键词
CMAC算法
机器人
神经网络
模糊控制
cerebellar model articulation controller (CMAC) algorithm
robots
neural network
fuzzy control
