Discussion on fault detection method for transmission line of LCC-MMC parallel three-terminal hybrid UHVDC system based on multi-layer perception

Since existing fault location methods suffer from the deficiencies of difficulty in selecting thresholds and the complex characteristic variables extraction，a fault detection method for line commutated converter-modular multilevel converter（LCC-MMC） parallel three-terminal hybrid ultra high voltage direct current （UHVDC） transmission lines based on multi-layer perception（MLP） is proposed. Firstly，the frequency characteristics of Kunbei-side boundary，T-zone boundary and Longmen-side boundary are analyzed，and the differences of fault characteristics when the fault occurs in different zones of the three-terminal DC line are point out. Secondly，the wavelet energy is extracted by multi-scale wavelet decomposition of line-mode current and line-mode voltage through wavelet transform，and the fault characteristic variables are formed combined with the positive- and negative-pole voltage variations. Then taking the fault characteristic variables as the input of MLP and the fault zone as the output，the fault zone identification model based on MLP is established. Then，fault zone can be identified by inputting fault characteristic variables extracted at measuring points to the trained model. A large number of simulations verify that the proposed method has high fault detection precision and can withstand a transition resistance of 300 Ω. © 2023 Electric Power Automation Equipment Press. All rights reserved.