Modelling of AC Transmission Line Faults Using Machine Learning Methods

Electrical transmission lines are a link between the generation station and the consumer. Any abnormality in transmission lines may result in a prolonged total electricity shutdown if not detected and identified accurately. Malfunctioning of transmission lines contributes to economic destruction and leads to wildfires and damage to electrical equipment. The existing literature has confirmed that 70 to 80 percent of the abnormalities that usually occur in transmission lines are single line-to-ground (L-G) faults. Detecting and identifying such faults will minimize restoration time. Machine learning-based models' accuracy has never been comprehensively tested in detecting and identifying L-G faults in transmission lines under noisy conditions. Noise in power system may arise from actions such as switching of transformers, circuit breakers, fluorescent lighting, voltage produced by poor connections, contact arching, voltage notching and the presence of surrounding environmental noise. This paper tested a MATLAB machine learning-based models to detect and identify L-G faults when introducing noise on these transmission lines. Simulation research data was used to validate the model functionality and accuracy.