Power load Forecasting based on VMD Multifrequency Combinations

Abstract

Power system load forecasting plays an important role in formulating the power system development planning, fuel planning and power generation planning. The traditional single model cannot fully characterize the fluctuating characteristics of load, scholars have attempted to build other prediction models based on empirical mode decomposition (EMD) or ensemble empirical mode decomposition (EEMD) to remedy this problem. However, the prediction accuracy of these prediction models is affected by modal aliasing and illusive components. Aimed at these defects, this paper proposes a multi-frequency combination prediction model based on variational mode decomposition (VMD), which improves deficiencies in prediction model based on EMD and EEMD. We use the back propagation neural network (BPNN), autoregressive moving average (ARMA) model, and least squares support vector machine (LS-SVM) to predict respectively high, intermediate, and low frequency components of a certain component. After obtaining the predicted values of each component, the BP neural network is applied to combine them into a final load prediction value. Finally, compare the prediction accuracy of the single prediction models (ARMA, BPNN, LS-SVM) and the decomposition prediction models (EMD and EEMD) with that of proposed VMD model according to the evaluation indexes such as mean absolute error, root mean square error and mean absolute percentage error, and we found that the prediction accuracy of the proposed VMD model is higher.