Abstract:
With the increasing electricity demand in Thailand to support the growing industrial, commercial and residential sectors, electricity demand forecast is crucial for electricity suppliers to manage the demand-supply chain to reduce the electricity cost as much as possible. The management is also known as Demand Side Management (DSM). For this paper, we present techniques for demand side forecasting using Support Vector Machine-Regression (SVM-R) with Radial Basis Function by analyzing one-minute intervals of electricity data collected from a sample group of industrials, commercials and residences. In spite of short-term forecast, we propose
the SVM-R model that forecasts a short-term load based on a previous week that performance measured by Mean Squared
Error (MSE), Squared Correlation (R2) and calculation time for forecast process. In addition, the model was tuned to optimize parameters based on real-life dataset collected from our research project. In the model, we consider variables including the time-of-day and the type-of-day such as workday, weekend or holiday.