Based on improved RBF neural network for chaotic time series prediction

Based on RBF neural network, making chaotic time series that was generated by Lorenz dynamical system as an object of study. The network prediction model was established. Input variables of network model have been optimized to improve. And compared to the BP, RBF neural network models, based on improved RBF neural network for chaotic time series forecasting model with higher predictive precision, smaller error and superior performance than the convectional BP or RBF neural network model, so the improved method is feasible and effective.