Estimation of Dominant Power Oscillation Mode using LSTM-RNN based on Synchrophasor Data

In recent years, power utilities are increasingly implemented variety type of generation source in the modern interconnected power system. So, major unexpected events affect an unstable of the power oscillation mode can cause a power system instability. The inter-area dominant mode estimation is very crucial for operation and prevention wide area blackout. This paper presents the Long Short Term Memory-Recurrent Neural Network (LSTM-RNN) approach for the estimation of an inter-area dominant mode based on synchrophasor data. Moreover, this approach provides an early critical warning during a modern power system takes a major disturbance. The critical index of the power system is considered at 5% of damping performance. Consequently, the critical early detection provides a great reminder for operators to determine intervention. The results show better accuracy of inter-area dominant mode estimation in comparison with Support Vector Regression- Polynomial (SVR-Poly) method.