Investigation and Improvisation of Fault Location Algorithm of Transmission Line Using Artificial Neural Network

Abstract

newlineAccurate fault location is important for maintaining reliability and efficiency of power transmission networks, particularly in high-voltage systems spanning over long distances. Transmission lines are the essential supporting structure of electrical power systems. They are responsible for transmitting electrical energy from the sources of power generation to distribution systems and ultimately to end users. Hence, it is of utmost significance to ensure that the entire power grid remains stable and operates effectively. This research focuses on investigating fault location techniques in high voltage transmission line network, addressing both symmetrical and unsymmetrical faults. The thesis proposes a novel approach for fault location and classification in transmission line by integrating Deep Neural Networks (DNNs) with effective signal analysis using Discrete Wavelet Transform (DWT) and Principal Component Analysis (PCA). DWT provides precise time-frequency analysis, enabling the detection of transient faults at the exact moment of occurrence. The model achieves 100% classification accuracy with perfect theoretical precision, demonstrating its ability to reliably distinguish between different fault types. Fault location performance is also significantly improved, with the model achieving the minimal mean square error (MSE) values, ranging from 5.008e-04 for Double Line to Ground (LLG) faults to1.1e-03 for Single Line to Ground (SL) faults. Results exhibit the effectiveness of proposed neural network approach in accurately locating faults along a model of transmission line rated for 132 kV spanning 100 km. Future research may focus on refining the proposed system under varying operational conditions of the transmission line. It may include optimizing the PCA and DWT parameters and thereby extending its application to different transmission line configurations and fault scenarios.