Abstract
Determining fault current direction in islanded inverter-based microgrids is highly challenging due to the unconventional and limited fault current contribution of inverters. This paper explores the application of a combined support vector machine (SVM) and synthetic harmonic injection method for determining fault current direction in such microgrids under balanced faults. A training set is created and an SVM is trained for each relay by applying forward and reverse faults in PSCAD/EMTDC and using MATLAB. The trained SVM is then used to predict fault current direction for a balanced fault at a different location. Simulation results are provided to analyze the performance of the proposed method.