Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A Survey of Open-Source Power System Dynamic Simulators with Grid-Forming Inverter for Machine Learning Applications

Tong Su, Jiangkai Peng, Alaa Selim, Junbo Zhao, Jin Tan

TL;DR

Open-source dynamic simulators are essential for ML studies in power systems because real-world data are scarce. The paper surveys five open-source dynamic simulators that support grid-forming inverters (ANDES, PSID.jl, Dynawo, OpenDSS, GridLAB-D) and analyzes their capabilities for TDS, GFM controls, and data generation for surrogate and control-oriented workflows. It includes an IEEE 14-bus case study showing close agreement between ANDES and PSID.jl under GFM controls, with differences arising from model choices. The results provide practical guidance for ML practitioners to generate representative datasets and select simulators aligned with modeling needs and computational constraints.

Abstract

The emergence of grid-forming (GFM) inverter technology and the increasing role of machine learning in power systems highlight the need for evaluating the latest dynamic simulators. Open-source simulators offer distinct advantages in this field, being both free and highly customizable, which makes them well-suited for scientific research and validation of the latest models and methods. This paper provides a comprehensive survey and comparison of the latest open-source simulators that support GFM, with a focus on their capabilities and performance in machine-learning applications.

A Survey of Open-Source Power System Dynamic Simulators with Grid-Forming Inverter for Machine Learning Applications

TL;DR

Open-source dynamic simulators are essential for ML studies in power systems because real-world data are scarce. The paper surveys five open-source dynamic simulators that support grid-forming inverters (ANDES, PSID.jl, Dynawo, OpenDSS, GridLAB-D) and analyzes their capabilities for TDS, GFM controls, and data generation for surrogate and control-oriented workflows. It includes an IEEE 14-bus case study showing close agreement between ANDES and PSID.jl under GFM controls, with differences arising from model choices. The results provide practical guidance for ML practitioners to generate representative datasets and select simulators aligned with modeling needs and computational constraints.

Abstract

The emergence of grid-forming (GFM) inverter technology and the increasing role of machine learning in power systems highlight the need for evaluating the latest dynamic simulators. Open-source simulators offer distinct advantages in this field, being both free and highly customizable, which makes them well-suited for scientific research and validation of the latest models and methods. This paper provides a comprehensive survey and comparison of the latest open-source simulators that support GFM, with a focus on their capabilities and performance in machine-learning applications.

Paper Structure

This paper contains 13 sections, 5 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Power System Dynamic Simulation and Machine Learning Applications
  3. Time-Domain Simulation
  4. Inverter-Based Resource
  5. Machine Learning Application
  6. Power System Dynamic Simulator
  7. ANDES
  8. PowerSimulationsDynamics.jl
  9. Dynawo
  10. OpenDSS
  11. GridLAB-D
  12. Simulation Comparison and Case Studies
  13. Conclusion

Figures (5)

  • Figure 1: Classification of power system TDS.
  • Figure 2: Structure of the GFM control: (a) Droop control. (b) VSM control.
  • Figure 3: Framework of interaction between DL, RL, and dynamic simulators.
  • Figure 4: ANDES simulation results: from top to bottom, the three rows correspond to G5 configured as SG, GFM-Droop, and GFM-VSM, respectively; From left to right are the speed $\omega$, active power generation $P$, and reactive power generation $Q$ of the SG or GFM, followed by the voltage $V$ of all buses.
  • Figure 5: PSID.jl simulation results: from top to bottom, the three rows correspond to G5 configured as SG, GFM-Droop, and GFM-VSM, respectively; From left to right are the speed $\omega$, active power generation $P$, and reactive power generation $Q$ of the SG or GFM, followed by the voltage $V$ of all buses.