LSTM-Based Fault Direction Estimation and Protection Coordination for Networked Distribution System

While the world’s power distribution system resembles an intricate web-like structure, most conventionally implemented mechanism is radial (RDS) where connection points for each line are normally kept open. However, disadvantages regarding traditional have led to active research on establishing networked (NDS), in which multiple circuits interconnected with electricity as well high-speed communication systems. The NDS offers advantages including increased facility utilization, hosting capacity, and higher terminal voltage. Conversely, unsolved issue prevails existing protection coordination method designed RDS inadequate fault occurrences due inaccurate direction identification. Hence, there urgent need alternative that allows high precision identification through communication. Moreover, application of various technologies hindered distance relay algorithm malfunctions situations lines short length, loads dispersed onto lines, integration distributed generation (DG) frequent. Therefore, this paper presents a uses waveform current based long short-term memory (LSTM) neural network communication-based scheme can be applied within NDS.