Study on Electricity Demand Forecasting Techniques and a Selection Strategy

Abstract

Based on extensive empirical studies, a taxonomy of well-known electrical demand forecasting approaches is presented in this study. In addition, a selection strategy for the approach of projecting electricity demand has been presented. The plan was developed with advice from experts in electrical demand forecasts and the World Bank's eight-factor model (through a questionnaire). On the basis of time horizon, accuracy, complexity, skill level, data volumes, geographic coverage, adaptability, and cost, the methodologies have been evaluated. The most widely used technique, according to the experts, is ARIMA (Autoregressive Integrated Moving Average) with exponential smoothing and Kalman filtering. Artificial Neural Networks with preprocessed Linear and Fuzzy inputs are the second most popular approach. Support Vector Regression, which is currently being examined by many electrical engineers involved in electricity demand forecasts, may now replace this approach. In addition to these methods that are highlighted, this study also provides ratings for alternative methodologies based on the World Bank's eight-factor model.  This research gives taxonomy of important electricity demand forecasting methodologies that will be helpful for practitioners, academics, and students. The report also includes expert opinions on the rankings of eight selection criteria based on a technique selection policy paper from the World Bank.