Based on approximation theory for multivariable functions, especially by neural network, determination of peak switching overvoltage by power systems has been described. A one hidden layer network trained by the backpropagation has been used to determine the number of hidden units, some heuristics are employed. The effectiveness and the accuracy of our approach are shown by comparing our results with that obtained by using the Electromagnetic Transient Program (EMTP)
Published in:
Neural Networks, 1995. Proceedings., IEEE International Conference on
(Volume:1
)
Date of Conference: Nov/Dec 1995