On Approximating the Dynamic Response of Synchronous Generators via Operator Learning: A Step Towards Building Deep Operator-based Power Grid Simulators
Researchers have developed a novel Operator Learning framework to approximate the dynamic behavior of synchronous generators, a crucial component in power grids. This framework utilizes Deep Operator Networks (DeepONets) to create neural network models that can either integrate with existing power grid simulators or act as a shadow for a generator's transient response. The approach includes a numerical scheme for simulating generator responses over time and a residual DeepONet that can incorporate existing mathematical models, complete with an error estimation. Additionally, a data aggregation strategy (DAgger) is proposed to fine-tune these networks for better performance during interactive simulations. AI