Bidirectional long-short-term memory-based fractional power system stabilizer: Design, simulation, and real-time validation

Power oscillations in modern power grids are inherent phenomena that may threaten system reliability. Therefore, to ensure acceptable system reliability, effective damping of power oscillations is inevitably required. In this context, this article introduces a novel approach to designing fractional power system stabilizer (FPSS) for effective damping of power oscillations. Bidirectional long-short-term memory (Bi-LSTM) approach is adopted to predict the parameters of FPSS. The conventional phase compensation technique is used to train Bi-LSTM network. To validate the efficacy of FPSS, different test scenarios of contingent operating conditions are simulated for the system. Comparative analysis is carried out with conventional power system stabilizers (PSSs) and optimization-based PSS techniques. Additionally, a test scenario is performed against existing deep neural network-based PSS methods to ascertain the robustness of the proposed PSS. Furthermore, the performance of the proposed Bi-LSTM-based FPSS is validated in real-time simulation using an interfaced OPAL-RT OP5700 hardware device.