Intelligent Fault Identification Framework for Power System Protection Using LSTM and Impedance-Based Fault Signatures

Electrical fault detection is one pf the most challenging condition in the power system. As observed in many power system analysis most of the time Single line to ground fault is the most common type of fault. Again many times the single line to ground fault requires low resistance fault. With the modern relay system, it is observed that most of the zone protection system requires exact evaluation of line impedance. Due to low resistance fault, eventually the fault current will be very less and the impedance seen by the relay will be closely associated with the actual resistance of the protected zones. Hence, the RoC will not perfectly match with the system fault impedance. In this paper ANN enabled fault detection technique has been implemented to identify the most three types of faults such as LG, LLG and LLLG fault detection. The research uses MATLAB as the platform to design and analyze the system.