A Novel Method for Detection and Location of Series Arc Fault for Non-Intrusive Load Monitoring

Series arc faults cause the majority of household fires involving electrical failures or malfunctions. Low-fault current amplitude is reason for difficulties faced in implementing effective detection systems. The paper presents a novel and faulty line identification method. It can be easily used low-voltage Alternate Current (AC) network Non-Intrusive Load Monitoring (NILM). Unlike existing methods, proposed approach exploits both voltage signal time domain analysis. Experiments have been conducted with up to six devices operating simultaneously same circuit an fault generator based on IEC 62606:2013 standard. Sixteen time-domain features were maximize arc-fault accuracy particular appliances. Performance random forest classifier was evaluated 28 sets five different sampling rates. For single period analysis arc, 98.38%, F-score 0.9870, while terms standard, it 99.07%, 0.9925. Location series (line selection) realized by identifying powered line. selection Mean Values Changes feature vector (MVC50), calculated absolute values differences between adjacent periods during fault. location 93.20% all cases 98.20% where affected device.