Enhanced Electricity Load Forecasting Based on an Improved Group Method of Data Handling with Fourier Residual Modification Approach

Abstract

Electricity load forecasting is crucial to modern energy management, planning and distribution in a highly developed country. Therefore, accurate electricity load prediction plays a vital role in maintaining optimal energy production and minimizing operational costs. This paper discussed an enhanced electricity load forecasting method based on a conventional statistical time series model with a combination method based on the Empirical Mode Decomposition (EMD) and Group Method of Data Handling (GMDH) model, namely EMD-GMDH. This study also presents a comparison between the proposed model of EMD-GMDH and a reconstruction of the Intrinsic Mode Function (IMFs) of decomposition integrated with Fourier residual modification, referred to as FM-EMD-GMDH. The Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) criteria were employed to measure the accuracy of each proposed model. From the empirical simulation, results reveal that the FM-EMD-GMDH model gained the best performance and better accuracy compared to other predictive models. Hence, it is concluded that the combination and reconstruction in the proposed model contribute to higher forecasting accuracy and performance.