Study on simulation of RBF NN identification method based on adaptive structural optimization

A novel nonlinear system identification method based on adaptive structural optimization of radial basis function neural network using particle swarm optimization algorithm is proposed in this paper. Using matrix encoding strategy, all parameters such as hidden layer nodes number, central position, directional width, weights of RBF NN are estimated dynamically in global. Under the framework of Structure Risk Minimization, the RBF NN model with excellent approximation ability can be dredged with prediction risk fitness. The simulation results show the effectiveness of this method.