Chaotic system identification based on Kalman filter

Presents a method to predict chaotic time series, including long-term prediction, and demonstrates the possibilities of constructing governing system equations based on the behavior of observed time series. First, a general system structure is assumed. Second, typical chaotic equations as reference system equations that show similar features with those of the observed time series are defined. Then the general system equations are approximated to system characteristics using a Kalman filter and attractors reconstructed in phase spaces. A sunspot chaotic system is taken as an example. Simulation results show that this method can identify the parameters of a chaotic system effectively and we construct a model which follows the Lyapunov uniform stability. The prediction of sunspot time series can get a high precision.