Approach for time series prediction based on empirical mode decomposition and extreme learning machine

Abstract

In order to improve the prediction accuracy of time series, a prediction method based on empirical mode decomposition and extreme learning machine is proposed. Through empirical mode decomposition, original time series can be decomposed into different frequency components. The components after decomposing remove the long correlation, prominent and the different local characteristics of time series, which can reduce the non-stationary of time series. Different extreme learning machine is introduced to prediction each component with different frequencies. The prediction value of each component is superimposed to obtain the final prediction result. Two typical time series include Lorenz chaotic time series and network traffic time series are chosen as the simulation object. Simulation results show that the prediction method in this paper has a good performance with smaller prediction error.