Day periodical classification for wide area day ahead short-term load forecast

Short-Term forecast technique is widely popular for accurate forecast in all sorts of future operation planning. In general future load is recognized as a non-linear mapping result from several previous step loads. This paper introduces a new ANN-based day ahead load forecast model for Wide Area in which loads are mapped from load pattern in previous day, rather than in previous steps load. With day periodical classification by k-means clustering, this new model achieves an excellent accuracy.