Prediction of Sub-synchronous Resonance Oscillations - A Machine Learning Approach

The use of series capacitors is an economical means to improve the load carrying capacity of long transmission lines, to enhance the transient stability, and to control load sharing be-tween parallel lines. However, this can cause Subsynchronous Resonance (SSR) frequency oscillations in the system resulting in shaft failures. In order to analyse SSR, the development of a complex mathematical model and small signal stability analysis are essential. This paper proposes a machine learning-based prediction of the possibility of unstable SSR oscillations and the torsional modes which can cause instability in the system for a given level of transmission line compensation. The test system considered in this study is the IEEE First Benchmark model proposed by the IEEE SSR task force. The proposed method is tested under different operating conditions. In most cases, the accuracy of the predictions is found to be sufficiently high (>90%).