Prediction about Chaotic Times Series of Natural Circulation Flow under Rolling Motion

The paper have proposed a chaotic time series prediction model, which combined phase space reconstruction with support vector machines. The model has been used to predict the coolant volume flow, in which a synchronous parameter optimization method was brought up based on particle swarm optimization algorithm, since the numerical value selection of related parameter was a key factor for the prediction precision. The average relative error of prediction values and actual observation values was 1.5% and relative precision was 0.9879. The result indicated that the model could apply for the natural circulation coolant volume flow prediction under rolling motion condition with high accuracy and robustness.