Input dimension reduction in neural network training-case study intransient stability assessment of large systems

摘要

The problem in modeling large systems by artificial neuralnetworks (ANN) is that the size of the input vector can becomeexcessively large. This condition can potentially increase thelikelihood of convergence problems for the training algorithm adopted.Besides, the memory requirement and the processing time also increase.This paper addresses the issue of ANN input dimension reduction. Twodifferent methods are discussed and compared for efficiency and accuracywhen applied to transient stability assessment