This study presents a method for detecting arc faults by combining load identification and MLP-SVM. The method addresses the issue of interfering loads on arc fault detection and the lack of significant arc fault features in some loads. Initially, the eigenvalues of the line currents for single and mixed loads are extracted in the time domain, both during arc fault and normal operation. Subsequently, load identification is performed using a complex matrix calculation method. After identification, an eigenmatrix and history matrix are created for each load.
To address the problem that the arc fault characteristics of certain loads are not significant and the line current characteristics cannot be detected effectively. In view of the wide application of a large number of nonlinear loads, it is difficult to distinguish the traditional current features from the load current features of nonlinear loads. Current features compared to those extracted from load end voltages, the latter are less affected by the load type, making the detection of faulty arcs more reliable.