摘要
针对自适应神经模糊推理(ANFIS)系统的结构辨识问题,提出了基于决策树的改进CART算法。改进算法引入了递归最小二乘估计器,对线性模型可降低计算量,并采用模糊技术处理不连续边界问题。由于隐含权值归一化,该算法能够快捷地对自适应神经模糊推理系统进行结构辨识。通过辨识仿真,表明了该技术辨识速度快,简捷方便,为ANFIS的结构辨识提供了行之有效的途径。
Improved CART algorithm of decision tree is put forward to solve at the problem of structure identification of ANFIS. Because RLSE is introduced, it can reduce calculation for the linear model. Besides, fuzzy technique is used to deal with the question of discrete boundary. Due to the normalization of connotative weights, the algorithm can identify the structure of ANFIS conveniently.The simulation result illustrates fast speed and convenience of the algorithm, and it provide an effective way for structure identification of ANFIS.
出处
《控制工程》
CSCD
2005年第S2期147-148,228,共3页
Control Engineering of China
