Identification of Power Quality Disturbance Using Neural Network

Electric power utilities have launched comprehensive data collection programs to evaluate the Power Quality (PQ) problems in their systems. A machine learning based classification and characterization of the power system data will help in dealing with the voluminous quantities of the monitored data. An Artificial Neural Network (ANN) based machine learning approach is developed for classifying the PQ disturbances in a power system. This paper illustrates how a pattern recognition neural network can classify power quality events on the voltage data at a 132 kV bus obtained from Maharashtra State Electricity Transmission Company Limited. Important measure of neural network performance is its ability to generalize, i.e. to respond properly to new data after training. It is tested on a set of new data through important performance measures like Mean Squared Error (MSE), Root Mean Squared Error (RMSE), correlation coefficient, regression analysis, error histogram, confusion matrix and Receiver Operating Characteristic (ROC) of the neural network. The results obtained demonstrate the power of ANN in classifying the commonly encountered disturbances in power system like sags, swells and interruptions.