Research on the application of RBF neural network based on K-means clustering in system identification

With a brief analysis of the strong points and drawbacks of RBF neural network, a RBF neural network based on K-means clustering algorithm is provided. The capability of nonlinear mapping and boundary distinguishing of RBF neural network together with the fast convergence of K-average clustering algorithm are both taken advantage of in nonlinear system identification. The simulation results indicate that the algorithm is fast to learn and precise to identify when this neural network is applied to nonlinear system identification.