Fault Location and Classification for MVDC Networks

This paper proposes a two step fault detection scheme for medium-voltage dc (MVDC) networks based on wavelet analysis. The proposed Grid Transients Classifier (GTC) is used to continuously monitor energy of wavelet packet decomposition levels (WPDL) for voltage and current signals to determine any transient disturbance like load change, pole to pole (PP) or pole to ground (PG) faults, pole to pole to ground (PPG), voltage sags, switching, etc. Subsequently, type of disturbance is identified through an artificial neural network (ANN) classifier. Then, proposed Active Grid Impedance Estimator (AGIE) injects a signal with two fundamental frequencies to evaluate grid impedance from the measurement point in new condition. Using the obtained data, fault type, location and its severity are specified. Hardware-in-the-Loop (HIL) results are provided for a shipboard grid case study to validate the real-time application of the proposed method and demonstrate its effectiveness.