非线性伺服机构的神经网络建模

Neural Network Modeling for Nonlinear Servomechanism

本文研究了利用径向基神经网络辨识某型伺服机构的问题。本文采用了具有自适应步长的Ｋ－均值聚类算法对网络中心进行调整，学习算法中的局部卡尔曼滤波器考虑了径向基网络的空间和参数局部特性，有效地提高了算法的收敛性与学习速度；

This paper discusses the nonlinear servomechanism modeling problem using the RBF neural network. We use the adaptive K means algorithm to adjust the neural network's center. By exploiting the spatially and parametrically local properties of the RBF network architecture and using the local extended Kalman filter, we proposed an computational efficient algorithm for training the network. The simulation results show that the approach is effective.