NEURAL NETWORK APPLICATION IN VOLTAGE HARMONICS PREDICTION

Existence of nonlinear element and also semiconductors in power system network, produce a large amount of harmonics. These harmonics which can cause misoperation of some devices such such as protective relays, filters, etc. and also failures of other ones like generators, motors, etc. need to be predicted in order to be eleminated. This paper uses, neural network, for predicting voltage harmonics in power system, and as these networks can be well used for modeling nonlinear, complicated relation, it is shown that in this case their application in highly accurate. The accuracy of this model is tested and verified for a sample data and it is shown that the mean error of this application for one hour prediction is so less that can be ignored.