Locating Faults on HVDC Transmission Systems Using Neural Network Model

The HVDC system is a viable option, particularly for long-distance transmission. Electrical transmission systems are susceptible to short circuit problems, resulting in excessive current that can ultimately damage the system’s equipment. Consequently, the system necessitates prompt repair to reinstate power transmission and ensure system safety. Therefore, the aim of this study is to create a model that can accurately determine the fault’s position. The objective of the effort is to develop a model that yields precise findings while also being collectively optimal. A bipolar transmission line, measuring 814 km in length, operates at 700 kV and has the capacity to transmit 1500 MW of electricity, constructed using PSCAD/EMTDC software in accordance with CIGRÉ benchmark requirements. The constructed model is subsequently simulated for a short circuit defect with an ON resistance of 0.01 Ω and an OFF resistance of 1.0 x 10^6 Ω, altering the fault location along the transmission line at 1 km intervals. The collected data was processed for feature extraction. Data from both terminals of the transmission line is utilized for the training and evaluation of deep neural network models. The assessment of the suggested system was conducted based on the mean squared error and the accuracy of fault estimation. The suggested system demonstrates superior performance compared to current baseline methods.