Automated Classification and Detection of Power Quality Disturbances using RBF Fault Classifier

Abstract

the proliferation of power electronic devices in a modern industrial control pronounced more power quality disturbances. There is an urgent need of technique which automatically classifies and detects power quality disturbances. In this paper authors developed an online radial-basis-function NN-based detection technique. In proposed scheme simple statistical parameters described which are used as input noise signals to classify vital conditions of power system like sag, swell of Induction motor, arc load, short circuit of welding machine, phase to earth fault and healthy condition. Detailed design procedure for RBF based classifier is presented for which experimental data of one HP, single phase, 50 Hz squirrel cage Induction motor, Welding machine to generate actual arcing load, Advantech data acquisition system is used. A Wavelet Transform Technique is applied to extract features from monitored data. By principle component analysis and sensitivity analysis dimension reduction is also achieved which classify the six types of PQ disturbances.