Control of One-arm Robot Based on RBF Neural Network Model

A method which is controlling of single arm robot is proposed that accorded to RBF neural network model.RBF neural network,who is named radial basis function, that has input layer, hidden layer,output layer,and it is a feedforward layer who has a single hidden layer. It is nonlinear that the input set mapped to the output set, but it is linear that the hidden layer space mapped to the output space. Because the RBF network whose action function is Gaussian basis function , its output is related to some modulation parameters. In the input space who is limited , it is not zero, and it is a neural network method who has local approximation. Therefore, RBF network can learn quickly and avoid local minimal problems, and the real-time requirements of control can be satisfy . Through establishing the Simulink control system simulation of the relevant model and the discrete digital simulation of M fuction, the control method which used RBF neural network model is very accurate and real-time, and the control effect is very good.