Electrical load forecasting using machine learning techniques

Abstract

Predicting the load in the power system for the next coming period is known as Forecasting, and the forecasting of the load from an hour to a day ahead is known as short-term load forecasting. Load forecasting is extremely useful for power generation, transmission, distribution companies, and independent system operators (ISO)s. Forecasting helps power utilities in maintaining their efficiency and revenues by assisting them in establishing supply-demand equilibrium. It also aids them in unit commitment, maintenance scheduling, future expansion, etc. newlineLoad forecasting is classified into three categories based on the time period for which it is forecasted such as short-term load forecasting ranging from a few hours to a week, midterm load forecasting ranging from a few weeks to 1 month and long-term load forecasting, which ranges from a few months to a few years. Short term load forecasting (STLF) improves automatic generation control performance and ensures revenue adequacy within the independent system operator (ISO). But the collection of noisy real- time data, complex design of power system, and extraction of various features found in load demand make accurate STLF modelling a challenging task. There always remains a research scope to develop a robust STLF scheme for precise estimation of future load demand patterns that helps the system maintain area generation control and resource dispatch. newline