A novel protection scheme for MMC-MTDC systems based on traveling wave characteristics and improved EEMD analysis

• Frequency-Based Guidance for EEMD Parameter Tuning: Provides quantitative conditions for selecting optimal decomposition parameters for extracting high-frequency fault transients based on early traveling wave propagation. • Improved EEMD for High-Frequency Transient Extraction: Develops a noise-controlled EEMD algorithm that enhances decomposition sensitivity and reduces mode mixing, enabling accurate isolation of early traveling waves. • Fast Local Protection Without Communication: Utilizes IMF 1 differential, IMF 1 energy, and modal voltages to achieve sub-2 ms protection without requiring communication between terminals. • High Robustness to Fault Conditions: Demonstrates reliable protection under varying fault resistances (up to 200 Ω) and noise levels (up to 30 dB), ensuring effective fault detection in challenging environments. The Modular Multilevel Converter based Multi-Terminal Direct Current (MMC-MTDC) system is a promising solution for long-distance, high-capacity power transmission, but its complex topology and high sensitivity to DC faults pose challenges for protection design. This paper proposes a novel protection scheme for MMC-MTDC systems that utilizes early traveling wave features extracted through an improved ensemble empirical mode decomposition (EEMD) algorithm. The method enhances the first traveling-wave front using band-limited noise injection, adaptive noise scaling, and intrinsic mode function (IMF) alignment with ensemble averaging, improving the detection of high-frequency transients. The protection scheme integrates a starting criterion based on the first intrinsic mode function (IMF 1 ) differential, fault identification using IMF 1 energy with a three-consecutive-samples condition, and pole discrimination through modal voltage analysis. Simulation results in PSCAD/EMTDC confirm response times under 2 ms, accurate detection with high fault resistance up to 200 Ω, and robust operation under 30 dB noise. Comparative analysis shows that the proposed method outperforms traditional EEMD and wavelet-based methods in terms of robustness and selectivity. The proposed scheme offers rapid, sensitive, and secure fault detection, making it suitable for the stringent protection requirements of MMC-MTDC systems.