Wind Power Forecasting in Short-Term using Fuzzy K-Means Clustering and Neural Network

Abstract

Wind power forecasting is the large emerging field in research as it plays a vital role in wind power plant operation. Wind power is one of the fast-increasing sustainable power resources and it can be viewed as the additional substitute for traditional power produced from non-renewable energy. Wind power forecasting can reduce over-dependence on traditional source of electricity. Due to the random behaviour of the airstream, there will discontinuity in collecting of the wind data which is the major impact for forecasting accuracy. Last decade many researchers have applied data mining technique in different prediction system that produced good accuracy. So, this paper proposed, a hybrid method consist of Fuzzy K-Means clustering and Neural Network(NN) are used to improve the forecasting accuracy and also to reduce computational complexity for forecasting the wind power in short-term. Fuzzy K-Means clustering is used for selecting similar days and it consisting of information about the weather condition and historical power data. To avoid the volatility problems, a backpropagation algorithm is incorporated into the NN. In order to prove this efficiency, a hybrid approach can be evaluated in actual wind farm which can give better forecasting accuracy and also expected to reduce computational complexity when compared with other existing wind power forecasting approaches.