A Combined Approach for Load Recognition and Arc-Fault Detection

Identifying connected loads in power grids is essential not only for smart power management but also to detect un-usual pattern indicative of faulty equipments or potential electric hazards. Among most dangerous electrical hazards, we find arc faults which sometimes may be identified as signatures of some common domestic loads. Thus, combining load recognition and arc fault detection in the same method is of utmost importance. In this work, we present a simple Self-Organizing Map (SOM) based method for load recognition and arc fault detection. The proposed method is trained on signals generated from real loads without and with arc faults. The obtained results show the effectiveness of the proposed approach in load recognition tasks and competitive results for arc fault detection.