Estimating electrical energy consumption using Linear Genetic Programming

Abstract

The electrical energy distribution subsystem is a component of the power delivery infrastructure that carries electricity from the high voltage transmission circuits to the customers. In order to prevent damage of the electrical energy distribution infrastructure, companies typically use transformers with digital meters that allow monitoring in real time, certain parameters like the amount of transformed energy. In several underdeveloped countries as Paraguay, the meters are installed only on a limited number of key transformers. Therefore, it is necessary to estimate the power consumption for the unmetered transformers using existing measurements. For this aim, this paper proposes the application of Linear Genetic Programming (LGP) to find good estimates of the power consumption of unmetered transformers. The proposal is compared with an analytical consumption estimation model proposed in a previous related work, being 13% better on average and 41% better in the best case. Dimensionality reduction proves to be useful to speed up calculation without losing much precision in the LGP estimations.