Intelligent Method for Fault Detection and Location in Transmission Lines
This work presents a method for fault detection and location in Transmission Lines (TLs). The proposed method is based on the extraction of Current Transient Deviation (CTD) curves, which are obtained in real-time and are used for establishing a threshold value used to detect faults. To perform the fault detection function, the RTDS will be used to generate fault signals and to execute the proposed method in real time, enabling a direct comparison with conventional digital relays. With respect to the fault location method, the CTD curves serve as input to a Long Short-Term Memory (LSTM) neural network, previously trained to estimate the fault distance. The methods were tested by using current signals acquired from the local terminal of a 230 kV and 100 km TL. Faults were simulated using different fault resistances, incidence angles and locations. The results demonstrate that the proposed methods can accurately detect and locate faults, regardless of fault characteristics and operating conditions. Additionally, a sensitivity analysis was conducted to evaluate the impact of each variable on the method’s performance.