一种新型模糊神经网络自适应控制器  

A New Neuro-fuzzy Network Adaptive Controller

作　　者：卢和[1] 王惠[1] 

机构地区：[1]天津理工学院

出　　处：《电气传动》2001年第4期31-33,共3页Electric Drive

基　　金：天津市自然科学基金资助 项目批准号 :99370 0 5 11

摘　　要：文章利用模糊神经网络的模糊推理能力以及前馈神经网络的逼近能力 ,将其与自适应控制方案结合 ,并取带有控制增量约束的广义目标函数作为优化指标 ;从而推导出一种能对非线性非最小相位系统进行有效控制的模糊神经网络间接自适应控制器。在网络学习算法上分别采用 Davidon最小二乘法和带有动量项的BP算法。仿真结果表明了该方法的有效性。In this paper,combining the adaptive control scheme with neuro fuzzy network which has the learning and adaptive ability of neural network and the human like reasoning and thinking flair of fuzzy system,a kind of adaptive control scheme is put formard based on neuro fuzzy network.The Back Propagation algorithm with momentum term is used in training the neuro fuzzy network as controller,and the Davidon Least Squares is employed in training the multi layer feedforward network as identifier.Through the simulation,the effectiveness and rapidity of the control scheme are demonstrated.

关 键 词：模糊神经网络 隶属度函数 自适应控制器 最小二乘法 

分 类 号：TP273.2[自动化与计算机技术—检测技术与自动化装置] TP183[自动化与计算机技术—控制科学与工程]