Short-term load forecasting in Singapore's energy market

Abstract

This paper presents a time series analysis for short-term electricity demand forecasting in Singapore. In the liberalised energy market, the Energy Market Company facilitates the wholesale market by providing market participants with price and energy demand forecasts at regular intervals. These forecasts help generators plan the amount of energy to produce ahead of the actual time period and ensure that the supply and demand of the grid are balanced. In this paper, deep learning models are implemented to improve the demand forecasts provided by the Energy Market Company. Particularly, 4 variations of Long Short-Term Memory models are implemented on Singapore’s historical load data from 2017 to 2020. The performances of these models are compared with the provided benchmark forecast.