A study on short-term electric load forecasting using wavelet transform

In this paper, we introduce data mining techniques for short-term load forecasting (STLF). The proposed approach is K-mean clustering and k-NN classification before forecasting. The K-mean algorithm to classify historical load data by season into four patterns. After that, the electric load which is clustered by seasonal patterns is divided into day-type pattern by use of the k-NN algorithm. The classified data are decomposed into an approximated with low frequencies and several detail parts associated with high frequencies. Each load component is fed as input to some forecasting models. We proved effectiveness using wavelet transform, the result showed high prediction accuracy.