Deep Learning Approach for Transmission Line Fault Classification

Abstract

As technology advanced, electrical interruption or disturbance still becomes a significant problem in power systems. A fault is one example of electrical disturbance or power failure in a power system. In order to recover the system, the fault must be detected, classify and locate to eliminate as fast as possible. Four types of fault occur in the transmission line. Those four types are Line-to-Ground Fault (L-G), Line-to-Line Fault (L-L), Double Line-to-Ground Fault (L-L-G), and Three Line Fault (L-L-L). These Days, fault has been one of the significant problems in the transmission line system. Fault can lead to power losses in transmission lines as well as power failure. Electrical service in the transmission line system needs to be recovered immediately after fault appears to avoid more energy losses. Thus, it is crucial to create a system that will detect and eliminate fault faster, more accurately, and effectively. Typically, transmission line fault classification required complex signal processing, required expert knowledge, and complex mathematical modeling to process the output signal. This paper proposed a deep learning technique to classify ten types of fault through simulation. The objective of this study is to propose automated signal processing and features extraction. This technique can model a system that generates the automated signal processing and extract features learning with a deep learning framework and classify all the ten fault types in transmission lines accurately and effectively.