Real-Time Classification of Excessive Neutral to Ground Voltage (NTGV) Using Support Vector Machine (SVM)

The excessive Neutral to ground voltage (NTGV) aggravates the operation of electrical system especially in communications, electrical appliance, and electronic data transfer. This corresponding problem contributes to the heating, negative sequence torque and the incorrect operation of the protection device. Thus, this study is focusing on developing the technique based on features extraction in real-time measurement in order to classify high NTGV. The objective is accomplished by developing the detection and classification system of high NTGV using S-transform, statistical analysis, and support vector machine (SVM). Further, the National Instrument (NI) voltage measurement module is utilized to acquire NTGV signal in real-time situation. In this case, the signal is generated using the AC Source Chroma Programming, where its signal is programmed according to the real data measurements in the distribution system. Next, the classification which will be done by using MATLAB software through Support Vector Machine (SVM) technique. This method is expected to enable the classification of different type of NTGV i.e harmonic, transient and combination of harmonic and transient. The result shows that the SVM technique produces high accuracy of classification, where its accuracy result is 95.8%.