Chaotic Analysis and Adding-weight Local-region Multi-step Forecasting Model for Wind Power Time Series

摘要

Wind energy is an inexhaustible energy and it has been widely used in wind power. However, wind power generation has a non-schedulable nature due to the uncertainty and randomicity of the wind speed. Hence, it is imperative to analyze the characteristic and prediction of wind power for power system operators. In this paper, wind power prediction is based on phase space reconstruction of chaotic time series modeling. Autocorrelation function and Cao method are applied to determine wind power time series of delayed time and embedding dimension, respectively. Then the positive largest Lyapunov exponent is obtained by using Wolf algorithm. The prediction of wind power time series from Adding-weight Local-region Multi-step Forecasting Model (AOLMM) is based on the largest Lyapunov exponent.