On Approximating the Dynamic Response of Synchronous Generators via Operator Learning: A Step Towards Building Deep Operator-based Power Grid Simulators

2023-01-29 · Source: cs.AI updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Utilities & Infrastructure · Depth: Expert, quick

Summary

A novel Operator Learning framework has been developed to approximate the dynamic response of synchronous generators, aiming to create deep operator-based power grid simulators. This framework utilizes a data-driven Deep Operator Network (DeepONet) to model the generators' infinite-dimensional solution operator. It includes a numerical scheme that recursively employs the trained DeepONet for simulating generator responses over time, interacting with multi-dimensional power grid inputs. Additionally, a residual DeepONet scheme is designed to integrate existing mathematical model information, accompanied by an estimate for cumulative prediction error. The framework also incorporates a data aggregation (DAgger) strategy for fine-tuning DeepONets with aggregated training data, ensuring robustness during interactive simulations. This approach effectively approximates the transient model of a synchronous generator as a proof of concept.

Key takeaway

For AI Scientists and Research Scientists developing advanced power grid simulation models, this operator learning framework offers a robust approach to accurately model synchronous generator dynamics. You should consider integrating DeepONet-based methods, especially the residual DeepONet for leveraging existing physics, to enhance simulation fidelity and predictive accuracy. This can significantly improve the realism and speed of your power grid transient response analyses.

Key insights

Operator Learning, specifically DeepONets, can accurately model synchronous generator dynamics for power grid simulation.

Principles

• Operator Learning approximates infinite-dimensional solution operators.
• Residual DeepONets can integrate existing mathematical models.
• Data aggregation improves DeepONet robustness in interactive simulations.

Method

Develop a data-driven DeepONet to approximate generator solution operators. Design a recursive numerical scheme for time-horizon simulation. Incorporate a residual DeepONet for model integration and error estimation. Implement a DAgger strategy for fine-tuning.

In practice

• Build neural network-based generator models.
• Shadow true generator transient responses.
• Enhance power grid simulator accuracy.

Topics

• Operator Learning
• DeepONet
• Synchronous Generators
• Power Grid Simulation
• Dynamical Systems
• DAgger Strategy

Best for: AI Scientist, Research Scientist

Related on AIssential

• Neural Operator-Based Surrogate Model for CFD:Helical Coil Steam Generator in Small Modular Reactor — Neural operator-based surrogates, combined with ROMs and multi-scale techniques, enable real-time CFD for SMR digital twins.
• Operator Learning for Smoothing and Forecasting — A new theory underpins data-driven smoothing and forecasting in dynamical systems using neural operators.
• Engineering Hybrid Physics-Informed Neural Networks for Next-Generation Electricity Systems: A State-of-the-Art Review — Physics-informed machine learning integrates domain physics into models, enhancing accuracy and efficiency for data-scarce electricity systems.
• LFNO: Bridging Laplace and Fourier via Transient-Steady Decomposition — LFNO unifies Laplace and Fourier operators to model dynamical systems across transient and steady-state regimes.
• Harness In-Context Operator Learning with Chain of Operators — Harnessing frozen neural operators with explicit transformations improves out-of-distribution generalization without fine-tuning.
• Reformulating Neural Operators in $d+1$ Dimensions for Embedding Evolution — Neural operators redefined in d+1 dimensions using Schrödingerisation better capture system evolution and achieve superior performance.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.AI updates on arXiv.org.