Enhancing fault detection and localization in MT-MVDC networks using advanced singular spectrum analysis

Abstract

This paper presents a novel methodology for fault detection, classification, and localization in Multi-Terminal Medium Voltage Direct Current (MT-MVDC) networks. The proposed approach utilizes Singular Spectrum Analysis (SSA) to decompose measured positive and negative pole voltages, isolating the seasonal component that represents the traveling wave. Fault detection is based on comparing this component against a predefined threshold, where minimal fluctuations occur under normal conditions, but significant variations emerge after a fault. Fault classification is achieved by analyzing the rate of change of the line-mode current to distinguish between forward and backward faults. For fault localization, the method leverages traveling wave attenuation and dispersion. The first traveling wave is extracted from the voltage seasonal component, and its spreading behavior over distance is analyzed to compute the curvature rate, enabling precise fault location estimation. The methodology is validated through extensive simulations on an MT-MVDC distribution system using PSCAD/EMTDC. MATLAB is employed for signal processing, and the approach is tested under various fault scenarios, including high fault impedance and extreme external faults. Comparative analysis with existing methods highlights the advantages of the proposed technique in terms of accuracy and robustness.