Generation system reliability evaluation using a cluster based load model

Abstract

A load model based on the concept of clustering is presented. Hourly loads are grouped on the basis of their proximity, resulting in nonequispaced load levels. For an equivalent accuracy, this load model uses much fewer states than conventional load models which have equispaced discrete steps. The reduction in the number of load states leads to more efficient calculations and a decrease in the computation time. A generalized convolution procedure is presented, and the convolution of the generation capacity model and cluster load model is shown as a special case of the general convolution procedure. >