Sheath fault detection and classification based on wavelet analysis

This paper presents a wavelet based algorithm to detect and classify sheath insulation faults on underground transmission cables. This algorithm can detect a high impedance sheath to earth fault, classify it, and discriminate it from other phenomena, such as load change, harmonics and surges on the cores. For the development of the algorithm, a simplified fault model is introduced to simulate a sheath to earth fault on a power transmission cable. Before applying wavelet analysis, an 'adaptive noise cancelling' (ANC) technique is used to extract the fault signal. Simulation studies are reported on a 300mm² concentric copper PVC insulated bonding lead.