Detection and classification of power quality disturbances using GWO ELM

Many industies have equipments sensitive to bad power quality that affects their production and product quality. Therefore, it is important to automatically monitor the quality of power with minimum human intervention. It is possible to analyze and interprete raw data from the industrial equipments to useful information with the help of signal processing and artificial intelligence system. This paper presents the automatic classification of power quality events using Extreme Learning Machine (ELM) in combination with optimization techniques. S transform is used for extraction of useful features of the disturbance signal. The features are used to train the ELM for classifying PQ events. Further the parameters of ELM are tuned through Grey Wolf Optimization (GWO) approach to improve the classification accuracy. Seventeen different categories of PQ events are used for the classification purpose. The efficiency of GWO-ELM is compared with other widely used classifiers such as K Nearest Neighbour (KNN), Support Vector Machine (SVM) and ELM. The simulation results reveal that the proposed approach can accurately detect and classify the PQ events.