Researchers from the Technical University of Denmark, including Ignasi Ventura Nadal, Mohammad Kazem Bakhshizadeh, Petros Aristidou, Nicolae Darii, Rahul Nellikkath, and Spyros Chatzivasileiadis, have developed a novel approach to accelerate electromagnetic transient (EMT) simulations in power systems. Their work, published in the IEEE Transactions on Power Systems, focuses on integrating Physics-Informed Neural Networks (PINNs) to enhance the efficiency of these critical simulations.
EMT simulations are essential for assessing the transient stability of power systems, particularly those with a high penetration of Inverter-Based Resources (IBRs) like wind turbines and solar inverters. These simulations are known for their accuracy but are often time-consuming, which can hinder real-time decision-making and system planning. The researchers identified the most computationally intensive components of EMT simulations and proposed replacing them with PINNs, which are neural networks trained to understand and replicate the underlying physics of the system.
The team demonstrated their approach using a type-4 wind turbine EMT model, achieving a 4-6 times speedup in simulation time by capturing the Phase-Locked Loop (PLL) with a PINN. The PLL is a critical component in IBRs that synchronizes the inverter with the grid. The researchers validated their results using PSCAD software, a widely-used tool for power system simulation.
The practical applications of this research for the energy sector are significant. Faster EMT simulations can lead to more efficient system planning, better integration of renewable energy sources, and improved real-time monitoring and control of power systems. This can ultimately contribute to a more stable and reliable grid, supporting the ongoing transition to cleaner energy sources. The modular and scalable nature of the proposed framework makes it adaptable to various components and systems, enhancing its utility across the energy industry.
This article is based on research available at arXiv.