Optimized Data-Driven Models for Short-Term Electricity Price Forecasting Based on Signal Decomposition and Clustering Techniques

In recent decades, the traditional monopolistic energy exchange market has been replaced by deregulated, competitive marketplaces in which electricity may be purchased and sold at prices like any other commodity. As a result, deregulation of industry produced demand for wholesale organized marketplaces. Price predictions, are primarily meant to establish clearing price, have become significant factor an company’s decision making strategic development. Recently, fast development deep learning algorithms, as well deployment front-end metaheuristic optimization approaches, resulted efficient enhanced prediction models that used price forecasting. this paper, six highly accurate, robust optimized data-driven forecasting conjunction with Variational Mode Decomposition method K-Means clustering algorithm short-term is proposed. work, we also Inverted Discrete Particle Swarm Optimization approach implemented method. The day-ahead based on historical weather data deregulated Greek market. resulting outcomes thoroughly compared order address two proposed divide-and-conquer preprocessing approaches results more accuracy concerning issue Finally, technique produces smallest error Decomposition, through variation Optimization, mean absolute percentage value 6.15%.