Comparative Study of Machine Learning Algorithms Applied to the Prediction of Unstable Sub-synchronous Resonance (SSR) Oscillations in Power Systems

Machine learning techniques have seen a surge in deployment in real-world scenarios due to ever-increasing availability of data and they have been successfully used in a large number of fields. The main aim of this project is the investigation of various machine learning algorithms and their application in a real-world engineering domain. The domain chosen is electrical power systems, which has seen increased use cases of machine learning. It is a dynamic and highly complex system that deals with the generation, transmission and distribution of electrical power. There is a need for high-speed control mechanisms that can quickly react to disturbances in the system to minimize the chances of failure. An important class of such disturbances is caused due to a phenomenon called Sub-synchronous Resonance (SSR). These are oscillations created due a resonance effect in the system and they can cause huge damage to the generator shaft. As a consequence, it may lead to the failure of the generator and could end up causing a complete power blackout. In this project, a machine-learning based approach is proposed that can predict the possibility of occurrence of unstable SSR oscillations, which can ensure that necessary precautions are taken to prevent shaft failures in the generator.