柔性直流配电网高阻接地故障检测方法

Flexible DC distribution network has acquired more interest and development in recent years, with the emergency of lots of renewable energy and direct current (DC) loads, as for the high impedance fault (HIF) detection in flexible DC distribution network, it not gathers attention from the industry and research community. Hence, this paper proposed an improved complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) method. Firstly, it utilizes CEEMDAN to extract the first intrinsic mode function component (IMF₁) of transient zero mode current (TZMC), and obtained the mutation singular point by calculating one order differences among IMF₁, then, it distinguishes fault stage from normal condition (NC) to calculate the slopes near the singular point, and compare slopes with start threshold. Secondly, uses Prony algorithm to identify the IMF₁ parameters, which including characteristic frequency component (CFC) and DC component, besides, calculates the energy ratio between CFC and DC to distinguish small impedance fault (SIF), medium impedance fault (MIF), high impedance fault (HIF) and load switching (LS). Lots of simulation experiments and field data prove that the paper can detect HIF effectively, and has some advantages compared with other methods in feature extraction, detection accuracy and calculation speed and so on.