This repository accompanies our paper Topology-Aware Reinforcement Learning for Tertiary Voltage Control submitted at PSCC 2024.

• The script that creates the three variants Standard, Condenser and Reduced from the initial case60nordic test case can be found here.
• The script that generates datasets from initial operating conditions can be found here.
• Links to the three generated datasets are provided below. Notice that it also contains the test sets modified by the baseline and the trained policies.
  • Standard dataset
  • Condenser dataset
  • Reduced dataset
• The tool we use to compare datasets can be found here

• The optimization baseline can be found here
• As this baseline only works with matpower data, one needs to preprocess (resp. postprocess) data before (resp. after) applying the baselin. Both scripts can be found here.

• The code used for the training of our Graph Neural Network policies is available here. Notice that this version provides trained policies as well as the configuration files that were used in our experiments.

A supplementary material to our paper can be found here.