Prediction of wind power by artificial intelligence techniques

The wind power generation in the Romanian power system will increase in the next years reaching almost 4000 MW by 2013. Taking into account the variability of electric power generated by wind power plants, the transactions on the balancing market will increase, leading to higher costs associated with the balance of generation and demand. It is our belief that using wind power forecasts can help in reducing these costs. This paper presents the results of a comparative study between two artificial intelligence based models applied in the specific case of predicting the total wind power installed in the Romanian power system - artificial neural networks and fuzzy inference systems. The tests show that both methods could deliver good results, provided they are used with sufficiently large training sets, but the fuzzy inference approach demonstrates better performances.