Fault detection method of flexible DC distribution network based on color relation analysis classifier

The flexible DC distribution network has the characteristics of low line loss, good power quality, fast system response, strong control and adjustment capabilities. It has become one of the mainstream trends in the development of the future energy internet. The effective detection of high impedance fault (HIF) is currently one of the key issues to be solved urgently in the flexible DC distribution network. For this reason, HIF detection method based on color relation analysis classifier (CRAC) is proposed. First, the complete ensemble empirical mode decomposition with adaptive noise algorithm is used to extract the intrinsic modal function (IMF) components. An IMF with the highest similarity is selected to calculate the IMF energy value in different states. Then, a starting threshold is set to distinguish between normal and abnormal states. At last, the CRAC is used to distinguish HIF, capacitor switching (CS), load switching (LS). Among them, the specific algorithm of CRAC includes the following steps: Firstly, the absolute value of the vector difference is obtained by subtracting the IMF components under normal and abnormal operation states. The absolute value is converted into Euclidean distance. Then, the Euclidean distance is transformed into gray grade. The mean value, maximum and minimum values of gray grade are converted into a red, green, and blue model. The model is transformed into a Hue-Saturation-Value color space model. At last, HIF, CS, and LS are distinguished according to the size of the hue angle. A large number of tests have verified the effectiveness of the proposed detection method.