Power System Synthetic Data Generation and Sharing

This panel focuses on power system synthetic data and topology generation and its various applications. Test systems populated by a large amount of realistic data sets are essential for the development, benchmarking, and evaluation of power system control and energy management algorithms. However, using genuine utility data sets as publicly available testing databases have been proven to be infeasible due to concerns involving data privacy and scarcity within the research domain. Nonetheless, to establish an equitable platform for algorithm performance comparisons, test beds with publicly accessible, highly authentic data sets and network topologies is imperative. We invited six panelists to present methods for synthetic data generation and its applications. First, an overview of the needs and limitations for generating and using synthetic data is presented. Next, different AI-based power system synthetic data generation methods for time-series profiles (load, wind, PV, etc.) and topologies will be presented. Metrics for evaluating the realism and limitations of each algorithm will be discussed. In the end, we will discuss the joint effort between the Technologies and Innovations (T&I) subcommittee and Big Data Analytics Subcommittee for sharing synthetic data generation tools and synthetic data sets. By hosting this panel, we hope to reach out to a broader audience and accelerate the establishment of a community wherein generative tools and the resultant synthetic datasets can be openly shared and evaluated using uniform performance metrics.