A Hybrid Approach for Improved Electric Short Term Load Forecasting

Abstract

For the daily management and planning of power grids, short-term load forecasting (STLF) is a crucial task in power systems. In STLF, the electrical load demand is forecasted from a few hours to several days in advance. For the power grid to remain stable and reliable, to avoid overloading or underutilizing power plants, and to maximize energy management tactics, accurate STLF is important. Various STLF methods, including those based on machine learning (ML) and artificial intelligence (AI), have been developed over the years, in addition to more conventional statistical techniques. Artificial neural networks (ANNs), support vector machines (SVMs), autoregressive integrated moving averages (ARIMA), and deep learning (DL) models are a few well-liked methods. Each technique has advantages and disadvantages, and the best one to use depends on a number of variables, including the availability of data, the time horizon for forecasting, and the level of accuracy required. The complexity and dynamic nature of power systems, the inherent uncertainty and variability in load demand, and the influence of outside factors like weather and human behavior mean that STLF remains a difficult task despite recent advancements. To increase STLF techniques' precision, robustness, and suitability for use in various power system scenarios, more research is required. Overall, STLF is a crucial task for ensuring the dependable and effective operation of power grids, and the development of precise and dependable STLF techniques is an ongoing research topic in the field of power systems engineering. Key Words: STLF: Short Term Load Forecasting, NN: Neural Network, RF: Random Forest, LR: Linear Regression,SVR: Support Vector Machine, Hybrid Meth