基于模糊神经网络的机械手自适应控制

单纯的神经网络和单纯的模糊系统具有各自的优点和缺点，模糊神经网络是两者的结合，它可吸取两者的优点而达到更优良的性能。这里提出了一种基于模糊神经网络的自适应控制方法。在利用常规控制器提取初始模糊规则的基础上，利用专家经验对初始规则进行补充，最后再利用误差的反向传播算法对参数进行在线的自适应调整。该方法用于机械手的跟随控制，两个模糊神经网络分别用于主回路控制和对象的逆模型，最后得到了优于样本控制器的跟踪控制效果。

Fuzzy neural network-based adaptive control of manipulators

Pure neural network and pure fuzzy system have advantages and shortcomings individually. The fuzzy neural network is the combination of neural network and fuzzy system, which gives better performance than any one of them. An adaptive control approach based on a fuzzy neural network was presented. First the initial fuzzy rules were extracted by using a traditional controller. Then expert experiences were used to complement the rules. Finally the error back propagation algorithm was used for on line adaptive tuning parameters of the fuzzy neural network. The presented approach has been used to a traching control of a manipulation by simulation. Two fuzzy neural networks have been used in the control. One has been used for the direct control, the other is used for inverse dynamics modelling. It shows a better result than the traditional controller.