DC line fault discrimination based on maximum information coefficient

Abstract

In DC networks, conventional line differential protection has the disadvantages of low reliability, weak resistance to transition resistance, and vulnerable to harmonic interference. In this paper, a fault identification method based on maximal information coefficient (MIC) is proposed for non-high resistance faults in the DC side of a flexible DC distribution network containing a modular multilevel converter (MMC) with single pole grounding DC protection inside and outside segments. The failure prediction is performed based on the reciprocal information at two sides of the line current. At the same time, for single-pole grounded high-resistance faults, the MIC is compared using the zero moduli of the current at two sides of the line for the recognition of high-resistance faults. The MMC-based flexible DC transmission network model is built and validated in the PSCAD/EMTDC platform. The results show that the method can identify faults protection inside and outside segments on the DC line area quickly and reliably. It also has the ability to resist harmonic interference and has an endurance of resistance for single-pole faults on the DC side; meanwhile, for high-resistance faults, it can also be detected accurately.