Slovenia has vastly expanded electricity production from renewable energy sources recently. In the renewables world solar energy prevails, because in last few years prices of photovoltaic modules have fallen steeply, which gives people extra motivation to invest in photovoltaic systems. Predicting electricity production from photovoltaic modules is very important for electricity distributors and traders in electric energy markets. We describe the Slovenian electric energy market with focus on daily products in which our predictive model can be applied.
In the thesis we show different approaches of processing raw data given from power plants and its visualisations, which are very important for easier understanding of the data. Attributes which are used for predictions are usually obtained in the form of weather forecast model parameters. Only the most valuable attributes are used in different machine learning models for predicting electricity production. Influence of spatial averaging multiple weather predictions for every power plant separately are studied, but our predictions are adjusted for Primorska region of Slovenia. We discuss and compare our results with other recent researches, where we reached a comparable or even better results.
|
