Intelligent Fault Detection and Location Method for DC Micro-Grid by Considering Different Fault and Non-Fault Operation Modes

In recent years, DC micro-grids (DCMGs) have become a subject of interest to researchers. One of the obstacles in the development of DCMGs is the protection of such micro-grids. Due to the fundamental differences between AC and DC micro-grids, the proposed protection methods for AC micro-grids are not suitable for DCMGs and therefore new protection methods should be provided. So far, many protection methods have been proposed for DCMGs, each with its own advantages and disadvantages. Due to the advances made in artificial intelligence, in this paper, a protection scheme for a ring DCMG based on machine learning algorithms and neural networks (NNs) is presented that is able to detect the occurrence of the fault and determine its location. In addition, the performance of the proposed protection scheme is evaluated in the event that one of the lines is out of service, which causes the micro-grid to get radial, and also in the event that the output power of one of the distributed generation sources (DGs) is reduced. In this paper, DIgSILENT software is used to simulate DCMG, and MATLAB software is used to train and test NNs and machine learning algorithms.