Chaotic Time Series Prediction Method Study

In the second half of the twentieth Century, along with the rapid development of nonlinear science, the research on chaos theory has become the main direction of non linear science research. Chaotic in nonlinear systems is a unique form of exercise, has now become a new science. Chaotic phenomenon exists in the material in any corner of the world, as large as the universe, as small as elementary particles, are subject to the chaos theory dominated. Such as:climate change, there will be chaos, chaotic phenomena also exist in the mathematics, physics, chemistry, biology, music, chaos broke the boundaries between different disciplines, is an emerging cross-disciplinary. It is both ubiquitous and extremely complex phenomenon, chaotic time series has also become an important application areas and research focus of the theory of chaotic prediction.In the study of chaotic systems, the need to determine the research system has chaotic characteristics. This paper first describes and discusses the chaotic characteristics of the system, the analysis of chaotic time series of noise smoothing method, and then introduced at this stage some of the commonly used chaotic time series prediction method and application of the phase space reconstruction method and gray theory to analyze the logistic model.Application of chaos theory of phase space reconstruction method and gray theory to analyze the logistic model, this paper presents a prediction method based on phase space reconstruction of chaos theory and gray theory:improved gray G M (1,1) model based on phase space reconstruction, the gray G M(1,1) model to improve model initial value correction, the initial value of adding a disturbance factor β, and application of the improved gray G M(1.1) model based on phase space reconstruction analysis model, the predictive value curve and error curve, and then predict the results with the prediction of the exponential model to obtain high prediction accuracy than the largest index prediction method of the Act, and to predict the measurement error is relatively stable.