Detection, Classification and Localization of Faults In Lvdc Microgrid Using Ann

Requirement of rapid fault clearance, which includes its detection, classification and localization, is the major barrier against secure implementation of DC microgrids. This paper presents a comprehensive protection method that uses the intelligent supervised learning algorithm to achieve not only successful fault detection in the DC microgrid but also classifies and localizes in a timely manner. The key feature of the proposed protection approach is the utilization of two local measurements i.e. voltage and currents, from the system for training the algorithm. The proposed approach utilizes conductance of the cable for detection of fault whereas voltage and current values are used for classification and localization of fault with improved accuracy. The proposed algorithm assists in achieving fast fault clearance within the acceptable time limits and accuracy. The performance of the proposed algorithm has been tested by generating various test cases including faulty and non-faulty conditions using PSCAD/ EMTDC software package.