Machine Learning Based Sub-Synchronous Oscillation Detection

Sub-synchronous resonance (SSR) is a problem that occurs in the modern electrical system (i.e. Generators with long-series compensated transmission lines) due to the Generator (Turbine) experiencing oscillations at different frequencies that are below their natural frequency. Various methods are used to compute the necessary conditions of SSR behavior and the eigenvalue analysis method is widely used. However, the conventional methods provide accuracy but they are not computationally efficient, and necessary steps cannot be taken immediately to mitigate the effects. This paper focuses on the implementation of a machine-learning approach for the detection of sub-synchronous resonance and its labeling.