Determining the load profiles of consumers based on fuzzy logic and probability neural networks

Abstract

Load profiles are a key issue in retail power markets where small consumers do not have the appropriate metering equipment. The main function of load profiles is in load balancing, and in the billing of consumers who have deviated from their contracted schedules. However, the act of establishing a system that enables consumers who do not have quarter-hourly load measurements to participate in the retail market, requires the determination of typical load profiles (TLPs) for various consumer types. In addition, a simple and straightforward method for assigning a TLP to a particular eligible consumer also needs to be established. In the paper a methodology for classifying consumers' load profiles is presented. The preprocessed measured load profiles (MLPs), using wavelet multiresolution analysis, were clustered with a fuzzy C-means clustering algorithm and an appropriate cluster-validity measure. A probability neural network was used to assign the TLP to a particular group of consumers. The results demonstrate the efficiency of the formation procedure for the proposed TLPs.