Data mining based transient stability prediction for power system integrated with varying levels of renewable penetration

The power system’s transient stability is affected by the increase in generation from renewable energy sources (RES). Applications such as the transient stability predictor assist the operator in initiating emergency control actions. In this work, predictors such as relative center of active power speed reference (RCOAPSR), relative center of reactive power voltage reference (RCORPVR), and inertia distribution index (IDI) are derived from the phasor measurement unit (PMU) measurements and fed to data mining models such as decision tree (DT) and random forest (RF). The aforementioned parameter’s area-wise calculations provide insights into the regional dynamics, which are important for RES-dominated systems. The simulations are carried out on the IEEE-39 bus system and northern regional power grid (NRPG) using DSA Tools software. The DT and RF models show a prediction accuracy of more than 97 % and 99 % respectively. The suggested scheme has an average early prediction time of more than 18 s. This can help prevent large-scale disturbance propagation.