Enhanced Fault Detection and Localization Strategy for High-Speed Protection in Medium-Voltage DC Distribution Networks Using Extended Kalman Filtering Algorithm

Conventional strategies are not effective in addressing the complex protection challenges in medium-voltage DC distribution networks (MVDCDN). The main challenge in MVDCDN is the high-rising DC fault current, requiring a robust and fast protection strategy. This paper proposes the use of an Extended Kalman filter (EKF) to detect various types of DC faults using only the current signal in the MVDCDN. In the first stage, current signals from the positive and negative poles corresponding bus are obtained. The EKF is then applied to the measured DC-current signals to generate two fault detection indices. The first index is the cumulative residuals (CR), calculated using the EKF iterative differencing process with updated current estimated state and noisy measurement. The second index is the modified DC version of total harmonic distortion, known as DC distortion factor (DCDF). The fault classification/zone identification (FCZI) unit is activated if changes in CR and DCDF are detected within the observation window of the relay. In the second stage, the FCZI unit calculates the Extended Kalman filter-based predicted energy (EKFBPE) for the faulty DC line section at both ends. The polarity of EKFBPE is used for fault classification and localization decisions. The proposed protection strategy requires low-band wireless communication capability in the smart grid. Extensive simulations using MATLAB Ⓡ Simulink 2022b are conducted on a ±2.5 kV MVDCDN with three feeders, considering various fault scenarios. The results demonstrate that the proposed scheme achieves 99.9% accuracy, under radial, looped, and meshed topology and is highly resilient to different types of faults with time of operation 1 msec. The scalability of proposed method and its effectiveness in handling higher voltage levels and associated fault uncertainty will investigate in future research.