Safe Bayesian Optimization for Data-Driven Power Electronics Control Design in Microgrids: From Simulations to Real-World Experiments

Daniel Weber,Eyke Hullermeier,Oliver Wallscheid,Henrik Bode,Jarren H Lange,Stefan Heid

doi:10.1109/access.2021.3062144

Abstract

Micro- and smart grids (MSG) play an important role both for integrating renewable energy sources in electricity grids and for providing power supply in remote areas. Modern MSGs are largely driven by power electronic converters due to their high efficiency and flexibility. Controlling MSGs is a challenging task due to requirements of power availability, safety and voltage quality within a wide range of different MSG topologies resulting in a demand for comprehensive testing of new control concepts during their development phase. This applies, in particular, to data-driven control approaches such as reinforcement learning, of which the stability and operating behavior can hardly be evaluated on an analytical basis. Therefore, the OpenModelica Microgrid Gym (OMG) package, an open-source software toolbox for the simulation and control optimization of MSGs, is proposed. It is capable of modeling and simulating arbitrary MSG topologies and offers a Python-based interface for plug & play controller testing. In particular, the standardized OpenAI Gym interface allows for easy data-driven control optimization. The usage and benefits of OMG for designing and testing data-driven controllers are demonstrated utilizing Bayesian optimization. Both the current and voltage control loops of a voltage source inverter operating in standalone, grid-forming mode for a remote MSG are automatically tuned given an uncertain application environment. Finally, the transfer to real-world laboratory experiments is successfully demonstrated.

Highlights

The transition of conventional energy supply systems based on fossil fuels to sustainable energy networks characterized by renewable energies is a central technical and social challenge of the 21st century [1]
CONTRIBUTION We present the OpenModelica Microgrid Gym (OMG) package, an open-source software toolbox for the simulation and control optimization of micro- and smart grids (MSG) based on energy conversion by power electronic converters [20], [21]
The core feature is a customizable interface between OpenModelica for plug and play-like system modeling and Python for the integration of arbitrary control algorithms

Summary

INTRODUCTION

The transition of conventional energy supply systems based on fossil fuels to sustainable energy networks characterized by renewable energies is a central technical and social challenge of the 21st century [1]. A. CONTRIBUTION We present the OpenModelica Microgrid Gym (OMG) package, an open-source software toolbox for the simulation and control optimization of MSGs based on energy conversion by power electronic converters [20], [21]. MICROGRID MODELICA LIBRARY Together with the OMG Python package, an OpenModelica library to create customized MSG topologies is provided [38] It mainly consists of freely connectable components like inverters, filters, and loads. The current and voltage control loops of a single VSI operating in islanded mode are investigated under uncertain operation conditions To this end, we will briefly summarize the fundamentals of the VSI system model followed by the inner level current and voltage control architecture, before linking this with a data-driven controller optimization. The entire following example can be found as executable code in the OMG repository [20]

BASIC SYSTEM MODEL

SAFE BAYESIAN OPTIMIZATION FOR CONTROLLER TUNING

SafeOpt

BASIC SETUP FOR CURRENT CONTROLLER OPTIMIZATION

Findings

CONCLUSION AND OUTLOOK