Research of low-voltage arc fault classification based on support vector machine

Support vector machine is introduced into low-voltage arc fault research, carrying out arc fault classification analysis for different loads. Referring to U.S. UL1699 standard, current data is collected by doing experiment. Then support vector machine is used to classify and compare the recognition results for different kernels. Light bulbs, switching power and cleaner are selected as typical loads to analyze arc fault. The differences between normal arc and arc fault are compared. Finally, it is concluded that arc fault current is usually short-time zero, has big differences between positive and negative half-cycle, and has large variation of amplitude and other characteristics. This provides reference for further arc fault research.