Bidirectional Gated Recurrent Unit Based-Grey Wolf Optimizer for Interval Prediction of Voltage Margin Stability Index in Power Systems

Despite the importance of Photovoltaic (PV) generation systems for smart grids, the uncertainty of PV power outputs due to weather variations can cause power stability problems leading up to voltage collapse in power systems. Governed by load dynamics, data-driven short-term Voltage Stability Margin (VSM) assessment is a critical technique for power system design and operation. This paper addresses improving the VSM accuracy and quantifying the uncertainties associated with the predicted values using deep learning. Interval estimation of VSM is significant for monitoring short-term voltage stability in real-time against fast voltage collapse and sustained low voltage without recovery. The proposed solution lies in the use of a Bidirectional Gated Recurrent Unit (GRU) optimized by a multi-objective Grey Wolf optimizer (GWO) algorithm. Time domain simulation results are conducted using IEEE59 and IEEE Nordic-44 bus systems to provide the training samples and maintain a high level of grid observability for the grid stability evaluation. The generated N-l contingency test cases data were conducted using Power System Simulator for Engineering (PSS/E)-based time domain simulations. The generated features from fault-induced voltage events include the post-disturbance voltage magnitude, angles, frequencies, and active and reactive power trajectories of the system buses. The obtained results show that the proposed method can enable timely stability assessment with higher effectiveness.