Load profile determination with artificial evolution

Load profiles are designed to approximate the average load curve of a certain class of end users. They can be used for commercial purposes as well as for load curve estimation. Load profiles are often very inaccurate as they do not take into account factors such as the type of housing of the end users or the presence of electrical heating. In this paper we present a method to determine accurate load profiles by handling the problem as a blind source separation, solved with a genetic algorithm. Data concerning load curves of 20kV feeders as well as a history of energy consumptions of more than 400,000 end users was provided by “É lectricité de Strasbourg Réseaux”. The load profiles found show considerable improvement in the estimation of 20kV feeder load curves.