基于改进最近邻聚类的机械手神经网络逆控制

机械手具有非线性时变、多变量、强耦合的特性，在机械手系统可逆的基础上，设计一种机械手的神经网络逆控制方案。通过神经网络逆辨识建立机械手的神经网络逆模型，把神经网络逆模型作为控制器模型与原机械手串联，构成一个伪线性动态模型，把非线性问题转化为线性问题。其中，辨识器和控制器均采用RBF神经网络结构，网络学习采用具有在线学习功能的最近邻聚类学习算法。仿真结果验证了本方案的有效性和可行性。

Neural Network Inverse Control of Manipulator Based on Nearest Neighbor Clustering Algorithm

Based on the reversibility of manipulator, a neural network inverse control strategy of manipulator was designed, which was in accordance with the characteristics of manipulator system such as highly nonlinear, time-variant, multivariable and strong coupling. The neural network inverse system of manipulator was established by neural network on-line identification, the model of controller which was the copy of identifier and the manipulator system in series, which forms a dynamic pseudo linear system, the control problem of non-linear plant is conversed into linear plant. The identifier and controller were based on RBF neural network which applied on-line learning nearest neighbor clustering algorithm. The simulation results show the validity and feasibility of the scheme.