Many typical faults in the distribution system, such as LG, LLG, LLLG, and LLL, cannot be recognized by standard protection relays because the fault impedance is too excessive. Several approaches for overcoming these issues have been described in the literature. Nevertheless, these approaches were incapable of classifying fault types or determining their severity. In this paper, a fault classification prevention approach for AC distribution systems is proposed using Variational mode decomposition (VMD) and Artificial neural network (ANN) techniques. The proposed technique has been tested against various defects such as LG, LLG, LLLG, and LLL situations. The precision is tested on an IEEE 13 bus system.