Contribution to the Management of Microgrids by the Application of Short Term Photovoltaic Power Forecasting

Abstract

Currently, smart MICROGRIDs or MICROGRIDs also called mini Smart grids permeate the whole world by their advantages which reside in a high degree of intelligence, autonomy and controllability. Therefore they require effective management within the different architectures of MICROGRIDs. Otherwise, these intelligent MICROGRIDS are made up of several microsources, namely one to several micro-turbines, a mini-hydraulic, one to several generating sets and renewable sources, namely wind and solar photovoltaic. The latter two sources are random and stochastic in nature and conditioned by meteorological parameters and are also trapped in climate changes during the days of the year. Indeed, the variability of meteorological parameters actually impacts the forecasting of the energy produced by natural resources (sun and wind). In short, the forecasting of solar energy becomes crucial to ensure the stability of the network and to allow an optimal unit commitment and an economic distribution. The objective of this study is to contribute to the modeling and improvement of the quality of short-term forecasting of solar photovoltaic energy in MICROGRIDs through the application of classification techniques. The classification technique envisaged for this work is the KNN model. To analyze results we compare the last technique with the persistence model. In order to test the performance of the different forecasting models, we used nMRE and nRMSE to analyze the performance aid criteria.