基于矢量自回归模型的循环平稳混沌时间序列的造型和预测.pdfVIP

Modeling and Prediction of Cyclostationary Chaotic Time Series Using Vector Autoregressive Models 1 Zhong Liu and Feng Xi Department of Electronic Engineering Nanjing University of Science and Technology Nanjing, Jiangsu 210094 Abstract It has been shown that some chaotic time series has cyclostationary characteristic. In this paper, this characteristic is exploited for applications to modeling and prediction of chaotic time series. To this aim, a vector-autoregressive-model-based model is developed. The model first transforms the scalar chaotic time series into a vector time series based on Wold’s isomorphism, and then uses the vector autoregressive model for modeling and prediction purposes. The application of the proposed model to simulated data from the periodically perturbed Logistic map is carried out and the results show that the model works well for modeling and long-term prediction in comparison with other models. Keywords: cyclostationary, chaotic time series, Vector Autoregressive model, Wold’s isomorphism I Introduction Signals, generated from many practical systems including physical, biological, physiological, economic, technological, and sociological systems, show chaotic behaviors due to inherent nonlinear characteristics of physical systems. The reconstruction of the generating systems becomes very important for understanding and processing