Discrimination of Power Quality Distorted Signals Based on Time-Frequency Analysis and Probabilistic Neural Network

Due to extensive utilization of sensitive devices, power quality issue has become moreimportant than before. so, accurate recognition and classification of power quality distortedsignals (pqdss) is an essential task in the power systems. in this paper two well-known timefrequency analyzers i.e. multi resolution analysis (mra) and generalized s-transfrm (gst)are applied simultaneously for extracting of some potential features. in order to choose thebest subset features, orthogonal forward selection (ofs) is used. ofs can rank featuresbased on their severability. probabilistic neural network (pnn) is considered as a powerfulclassifier core for discrimination of dominant selected features. extensive samples of pqdss aresimulated to evaluate the performance of the suggested detection scheme. also, sensitivity of theproposed method has been investigated under different noisy conditions. at last the obtained results are compared with the accuracies of some reported methods of previous researches.