CAREER: Transforming Distribution System Situational Awareness via Continuous-Time Adaptive Data Fusion

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2).

In the path to the clean energy future of the nation, the electric power industry is witnessing a massive shift from centralized fossil fuel generation to distributed renewable energy generation. Power distribution systems are responsible for connecting customers to the main grid and balancing the instantaneous generation-load mismatch at each customer.

As power flows in the distribution systems become highly volatile and bidirectional, it is of crucial importance to gain situational awareness in real time such that grid operators can assess and enhance the renewable energy hosting capacity, and customers can reliably, resiliently, and efficiently buy/sell electricity to meet their needs. As a variety of data sources are being populated in distribution systems, the fundamental question remains how to extract and integrate the information to construct a complete picture of distribution system operation.

Existing methods have not fully considered the complicated measurement environment pertaining to power distribution systems, and cannot produce accurate and reliable results when the measurements have diverse sampling rates, sampling times, accuracy classes, and with limited communication support. This project will develop transformative concepts and methodologies to comprehensively address the outstanding challenges in tracking the operating states of distribution systems.

The outcomes of the project will fully bring out the potential of various sensor assets for distribution system situational awareness, which will serve as the foundation of intelligent decision making processes for accommodating the volatile but pervasive distributed renewable energy generation across the grid. The project will feature a Seeable Electrical Energy Distribution (SEED) program to integrate research with education.

It will develop a simulation and visualization platform for a close-to-real synthetic distribution system “operating” in real time 24/7, providing educational experience to the wide public that has not been possible without entering control rooms of a utility company. The platform will also serve as a public data portal for researchers around the nation, facilitating data availability and research reproducibility across the whole technical community.

State estimation is a key technology for enabling the situational awareness of distribution systems and massive integration of distributed renewable energy generation. The existing distribution system state estimation methods largely inherit mature concepts from state estimation of high-voltage transmission systems, and do not fully consider or address the unique complicated measurement environment in distribution systems, including the unknown continuous-time state transition model, asynchronous and multi-rate measurements, unknown and time-varying measurement error statistics, and limited sampling rates and communication bandwidth.

This project will propose a revolutionary distribution system state estimation paradigm that will transform the situational awareness of distribution systems for accommodating massive and pervasive renewable energy integration and demand response. We will develop new concepts and methodologies that result in a holistic solution allowing 1) learning-based continuous-time system dynamics modeling, 2) seamless fusion of asynchronous and multi-rate measurements arriving at any continuous time instants, 3) adaptive near-optimal estimation under unknown and time-varying measurement error statistics, and 4) proactive scheduling of sensor sampling times to maximize observability and minimize communication congestion using clustering.

With the continuous-time data fusion feature, the proposed paradigm will replace the conventional discrete-time step-by-step estimation paradigm and reshape the field of distribution system state estimation. In a unique Seeable Electrical Energy Distribution (SEED) program, generative adversarial network will be exploited to synthesize distributed renewable energy and load data, which cannot be distinguished from real-world data yet do not have proprietary issues and can be freely distributed and reused by the research community.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.