LSTM Based Real-Time Transient Stability Assessment Using Synchrophasors

This paper proposes a data driven model based on long short-term memory neural network that identifies the transient stability status of power systems using synchrophasor measurements. The proposed method relies directly on synchrophasor measurements unlike the existing work which rely on the rotor angle estimation. Therefore, it bypasses the errors encountered during the rotor angle estimation. The measurements from the phasor measurement units are dimensionally reduced to obtain the sequential input data to train the long short-term memory model. The proposed method performs faster than the existing work reported in the literature and is independent of the information of an event’s time during online deployment The results are verified using synchrophasor measurements obtained for an IEEE 39 bus system and implemented as a software-in- loop approach using a real-time digital simulator to verify its deployment within a control room.