A Time Series Prediction Method Based on a Modified Radial Basis Function

Time series analysis develops models that can establish the relationship between different variables. For nonlinear systems using time series analysis we propose to combine the 4 techniques of: i) Radial Basis Function (RBF), ii) artificial neural networks, iii) adaptive control and iv) optimization, and explore the design of robust control algorithms for uncertain nonlinear systems. Then, based on an improved RBF, the problems of prediction models using time series are investigated and lead to the design of a nonlinear system adaptive output feedback control algorithm without the need to add the state observer in the system control. The experiment shows that the prediction result of the improved method is better than the usual RBF network, the optimization method can confirm the stability of system and speed up the convergence rate of nonlinear model of time series analysis, which achieved a much better modeling and prediction accuracy than other existing models.