Multi-model approach for electrical load forecasting

Electricity forecasting is a big deal for companies, and so the energy planning is needed in the short, medium and long term. In this way, it is important that the prediction remains relevant taking into account different parameters as GDP (Gross Domestic Product), weather, and so on. This work focuses on forecasting medium and long terms of Algerian electrical load using information from past consumption. This article uses time series models to forecast, different models have been implemented and tested on a database, which represents ten years of consumption. The studied model consists in predicting months and years using implicit information contained in historical ones. Three models are implemented in this work. Multiple linear regressions, artificial neural network MLP (multilayer perceptron), SVR (Support Vector Machines Regression), a parallel approach using seasons decomposition is used to have a more accurate result. One of these proposed models is relevant and is an encouraging forecasting model.