Multi-scale Integration of Physics-Based and Data-Driven Models in Power Systems

The major subject of this paper is the introduction and testing of a new modeling paradigm necessary for enabling sustainable performance of electric energy systems. In previous work we have identified major challenges to systematically modeling distributed, non-uniform resources emerging in power grids. Today´s modeling of electric energy systems is either entirely based on first principles which suffers significantly from the ever-increasing complexity of non-uniform devices, or is purely based on computer science data-driven approaches which lose the fundamental physical insights of electric power networks. Therefore, it is very difficult with today´s modeling practices to integrate distributed non-uniform resources using both first-principle and data driven approaches in large-scale cyber-physical energy systems. In sharp contrast, this paper presents a holistic multi-scale modeling approach by combing advances from (1) physics based modeling of emerging distributed resources (e.g. wind generation and storage devices), and (2) data-driven modeling of load resources. With both physics-based models of distributed resources and data-driven models of flexible demands, key parameters are abstracted and identified from the detailed dynamical models necessary for the multi-scale power system operations. The proposed modeling framework is tested using realistic phasor measurement unit data obtained from Electric Reliability Council of Texas (ERCOT).