Research on a Hybrid EMD-SVR Model for Time Series Prediction

Abstract

Time series prediction methods were widely used in various fields. The prediction method for non-stationary and nonlinear time series was studied in this paper. This method decomposed non-stationary time series into stationary sub-sequences using the Empirical Mode Decomposition method. And then an appropriate time-step was chosen and the Support Vector Regression algorithm was applied to predict each stationary sub-sequence. The sum of predicted values was the forecasting results of the original sequence. The method was applied to building energy consumption datasets, which were collected in some buildings. The experimental results showed that the hybrid algorithm of Support Vector Regression and Empirical Mode Decomposition had higher accuracy and was suitable for predicting non-linear and non-stationary time series. Moreover, this hybrid algorithm was used to predict the time series with outliers and to test its noise-resistant performance. The forecasting results also illustrated EMD-SVR algorithm was more robust than SVR algorithm.

Hybrid Time-Series Prediction Method Based on Entropy Fusion Feature

“…Te EMD decomposition of the correlation prediction model of the sequence data with time characteristics has been developed in recent years [32][33][34]. Zhang et al's [35] derived model of EMD decomposition improved the gating cycle unit (GRU) and combined EMD with the regression prediction model of PM2.5.…”