Yinyan Liu

Postdoctoral Research Associate

UNSW Sydney

Yinyan Liu received her B.E. degree in measuring & controlling technology and instrument from North China Electric Power University, China, her M.E. degree in control engineering from Tsinghua University, China, and her PhD degree in the School of Electrical and Information Engineering, from the University of Sydney, Australia in 201, 2018, and 2022, respectively. She is currently a Postdoctoral Research Associate at the School of Photovoltaic and Renewable Energy Engineering, University of New South Wales (UNSW), Australia. Her research interests include fault diagnosis of PV systems, energy management for decarbonisation and decentralisation, energy efficiency for consumers with distributed energy resources, pricing and scheduling of shared energy storage, and internet-of-things with the understanding of human factors.

Residential Energy Management for Renewable Energy Systems Incorporating Data-Driven Unravelling of User Behaviour

The penetration of distributed energy resources (DERs) such as photovoltaic (PV) at the residential level has increased rapidly over the past year. It will inevitably induce a paradigm shift in end-user and operations of local energy markets. Specifically, the local energy communities and their members can legally engage in energy generation, distribution, supply, consumption, storage, and sharing to increase levels of autonomy from the power grid, advance energy efficiency, reduce energy costs, and decrease carbon emissions. Non-intrusive load monitoring (NILM) can extract the users' load consumption from the smart meter to support the development of the smart grid for better energy management. Yet to date, how to design residential energy management, including home energy management systems (HEMS) and community energy management systems (CEMS), with an understanding of user preferences and willingness to participate in energy management, is still far from being fully investigated.

This research bridges the gap in residential energy management for RES by incorporating data-driven unravelling of user behaviour in the existing literature. The contributions of this thesis lie in three aspects: 1) This research develops several novel neural network models to remove the unrealistic assumptions of NILM, improve the generalization of the NILM algorithm, and reduce the dependency on the labelled data for unravelling users, 2) uncertainty scores of NILM results and quantitative user characterization are creatively incorporated into an advanced HEMS to schedule the appliances and respond to the demand response signals for economic benefits., and 3) a CEMS incorporating user clustering is designed to maximize the techno-economic-environment benefits.