Fault Detection and Classification in the HVDC System using Machine Learning Algorithms

There are many challenges are faced by the power system such as electrical load growth, axillary supply. Due to expansion, population, urbanization, and environmental concerns, it is become more and more difficult to meet power demand for new lines. In order to find the proper transmission system with low cost, high efficiency, less loss, etc. and to feed the power to increasing power electronics system in the advancing world, the innovators now got view of HVDC system which will eliminate the current concerns in the transmission. But the stepping into the new system will surely needs the study and analysis to find the possibilities of reliability and concern states in the real word implementation. One most important factor in such system is Fault analysis. Fault analysis in the HVDC system is not developed as compared to existing AC system. HVDC systems have many fault conditions such as pole-to-pole fault, positive to ground fault, negative to ground fault, and other minor and major faults. Here, this paper investigated various failure conditions that can occur in HVDC systems and provide the results of analyzing each failure condition of HVDC. Data for the various type of faults were extracted and that were trained and tested with the different machine learning classifier in MATLAB. Around 80,000 samples were trained. Form the results, the MEDIUM classifier got accuracy of 92.6% and COARSE classifier got accuracy of 89.6%. But FINE SVM classifier got a highest accuracy (96.9%).