Transmission loss allocation in a power market using artificial neural network

Abstract

An artificial neural network (ANN)-based solution of the transmission loss allocation problem in a power market is suggested. The ANN proposed in this paper is a multilayer Perceptron network based on Levenberg–Marquardt algorithm and is capable of allocating losses to the agents identified as transactions in a power market. The network has a dynamic composition in the sense that it has to be trained afresh for determining the loss allocation of every transaction scenario instead of a network which is trained only once for all possible scenarios. The training dataset required is only a few in numbers and is filtered out from a large pool of data. The data pool includes the transactions values and their corresponding allocation of losses computed according to some established allocation method. Performance of the NN following game theoretic and proportional allocation of losses has been reported. Results are produced on standard test systems for bilateral and pool market operations.