WSEAS Transactions on Power Systems
Print ISSN: 1790-5060, E-ISSN: 2224-350X
Volume 20, 2025
Adaptive Generator Tripping Scheme based on Deep Learning as Real Time Control Action for Transient Stability
Authors: , ,
Abstract: Transient stability has typically received attention in the literature, focusing on its assessment and control under limited operational scenarios and contingencies. This often results in persistent transient stability issues, leaving the system vulnerable to imminent collapses. In this regard, this work aims to develop an adaptable tripping scheme based on the power system dynamics following a major disturbance to prevent grid blackouts due to transient stability loss. The proposed methodology takes advantage of data analysis tools based on deep learning and Phasor Measurement Units (PMUs) technologies. In this approach, the methodology involves generating a database of both operational scenarios and n-1 contingencies, labeling critical generators to be tripped to mitigate transient instability, and training a hybrid deep neural network RCNN (recurrent convolutional neural network) that constitutes the core of the tripping scheme. Following the application of the methodology in a controlled simulation environment, the RCNN model demonstrated strong performance, as it not only mitigated transient instability through minimal tripping of generation plants with an effectiveness of 92.4% but also showed potential for real-time application, as the control action accounts for latencies inherent in real-time operation.
Search Articles
Keywords: Power systems, transient stability, deep learning, PMU, generator tripping scheme, critical generators
Pages: 1-13
DOI: 10.37394/232016.2025.20.1