DWT and Bayesian technique for enhancing earth fault protection in MV networks

In this paper, a Bayesian selectivity technique is introduced to identify the faulty feeder in compensated medium voltage (MV) networks. The proposed technique is based on a conditional probabilistic method applied on features extracted from the residual currents only using the discrete wavelet transform (DWT). DWT enhances to localize initial transients generated in the network due to the fault event. The absolute sum of a window of the DWT detail coefficient is used to detect the fault. The conditional probability provides the selectivity decision. The fault cases occurring at different locations in a compensated 20 kV network are simulated by ATP/EMTP concerning practical fault case such as arcing faults. Test results corroborate the efficacy of proposed technique.