Chaotic Time Series Gray Correlation Local Forecasting Method Based On Fractal Theory

Abstract

The fractal and the structure of self-similarity are two main features of chaotic behaviors. Based that an initial search for a self-similar trajectory, the series of gray correlation for local forecasting chaos time are proposed. The nearest neighbor points in a phase space are selected based on Euclidean distance and the degrees of gray correlation. The method proposed is applied to a short-term, power-load forecasting. The results of load forecasting using this approach indicate the method is efficient and accurate.