Reconstruction And Prediction Of Cyclostationary Chaotic Time Series

Chaos has acquired wide attention because of its rich dynamical behaviors. This thesis mainly investigates cyclostationary properties and prediction of chaotic signals generated by a periodically perturbed logistic map (PPLM).Based on cyclostationary signal theory, the cyclostationarity of the PPLM chaotic signals is firstly analyzed. Simulations show that the signals exhibit the cyclostationarity. Then, combining the cyclostationary properties with autoregressive prediction model, a periodic time-varying autoregressive model (PTVAR) for predicting the PPLM chaotic signals is developed. Finally, the prediction performance is compared with that by radial basis function network, and simulation shows the performance advantages of the proposed PTVAR model.