POWER LOSSES ESTIMATION IN REAL POWER DISTRIBUTION SYSTEMS BY MEANS FEED-FORWARD NEURAL NETWORK

Unlike test systems in which electrical power losses is calculated using straightforward and simple methods, this procedure faces with various challenges in realisticnetworks. Lack of quality and availability for input data are main roots of such challenges. In order to cope with this issue, this paper proposes a feed-forward neuralnetwork (FNN) for calculation of electrical energy losses in realistic distribution systems. In this regard, firstly, input parameters of the model are extracted taking intoaccount two factors namely ۱) mathematical formulation of power losses, and ۲) data availability in realistic power distribution systems in Iran. The input parametersand output parameter, i.e. electrical energy losses, are used to develop an FNN. Finally, realistic data from ۳۹ DISCOs in ۸-year time horizon are gathered, preparedand used to train the network and examine its performance. Simulation results demonstrate that the outputs of network significantly trace the pattern of lossesamong DISCOs. Moreover, losses amount is predicated with acceptable accuracy.

Keywords:

Distribution systems , electrical energy losses , feedforward networks , neural networks.