Arc fault identification method based on fractal theory and SVM

The effect is very large of arc fault in electric conductors to the users´ electric safety. The test system to arc fault was built based on UL1699 standard in order to detect and identify the arc fault in the electric wires which can collect the arc fault current in the typical electrical equipments. Phase space planar graph of the current waveform in time domain is obtained using the phase space reconstruction method, and geometric dimension and information dimension of current phase planar graph are calculated using fractal theory to build the sample database of current two dimensional characteristics value. Non-linear sorter system of the radial basis function was built based on vector machine theory to train and test the sample data of characteristics value. Simulation result of Matlab proved that this method has high distinguishing degree and can identify the arc fault in electric wires accurately and reliably.