摘要
对于强非线性系统采用分段建模十分有效,广义模糊神经网络能实现这种思想。在此基础上,给出一种模糊规则前、后件参数可分别进行学习的算法,仿真结果表明该方法拟合能力强、学习效率高。
A strategy was presented for approximating a function by combining fuzzy neural networks with piecewise models. A decoupled learning algorithm was given. The condition and consequence of fuzzy rules can be separately trained. Computer simulation shows that the algorithm is feasible and efficient.
出处
《控制与决策》
EI
CSCD
北大核心
1997年第5期622-624,共3页
Control and Decision
基金
广东省自然科学基金
国家攀登计划认知科学(神经网络)重大关键项目资助课题