A forecasting model for predicting electricity day-ahead market prices for scheduling industrial applications

Abstract

Over the last 20 years, electricity price forecasting (EPF) modeling has become a fundamental process of the decision support mechanisms of energy companies for successfully participating in the power markets. Given the growing uncertainty and the stochasticity of the evolving energy mix, this process has also proven to be of importance for industrial companies whose production scheduling decisions are greatly affected by the energy costs. While different approaches have been tested (with varying degrees of success), in this study the accurate Day Ahead Market (DAM) prices prediction, in different European markets, are directly associated with the production scheduling processes of various industrial loads. We propose a configuration based on a Random Forest (RF) model implementation for the prediction of DAM prices for a number of European countries with interconnected electric (market) systems. The proposed model for a particular country is enriched by including power market information from neighboring countries, which in turn consistently improves the accuracy of the forecast. The distributed developed DAM price forecaster allows industries to schedule optimally their production processes with respect to energy cost reduction and energy efficiency improvement. The optimality of operations, sometimes, requires rescheduling of production activities in neighboring countries in the case of industries with multi-site production facilities.