Coding
Loading selected year
Less
More
I am a PhD Candidate in the Institute for Environmental Design and Engineering, The Bartlett, University College London, supervised by Prof. Rui Tang and Prof. Ivan Korolija.
My research focuses on intelligent building control and building-grid interactions, with an emphasis on multi-building coordinated control for demand response, voltage regulation, and demand flexibility provision. I work on developing advanced and practical Reinforcement Learning algorithms, particularly safe RL, for modern building energy systems and power systems.
In parallel, I work on constructing co-simulation framework of simulation programs (e.g., EnergyPlus and Modelica) to build scalable and physics-based, open-source control testbeds that support RL implementations, control analysis and benchmarking for smart building/grid applications.
Loading selected year
-- basketball, -- strength, -- both
I attended the International Conference of Building and Simulation for Climate-Neutral Buildings and Cities, held from 11 to 13 June 2026 at The Hong Kong University of Science and Technology (HKUST), Hong Kong, China.
I presented my study, Multi-Building Coordinated Control of Grid Voltage on a Physics-Based Simulation Platform Using Constrained Reinforcement Learning. The work develops CSAC-VR, an advanced safe reinforcement learning algorithm for coordinating multiple buildings on a physics-based simulation platform, enabling building flexibility to support grid-voltage regulation while respecting operational safety constraints. It enables safe and efficient interaction and operation between buildings and the power grid. It received the Best Paper Award.
Beyond the presentation, I had the chance to meet and learn from many teachers and students from different universities. The conversations and new connections made the conference especially rewarding.
Poster available at: Poster
I developed a scalable, physics-based multi-building co-simulation platform that supports simulating building clusters using both EnergyPlus and Modelica, enabling RL training, evaluation, and reproducible benchmarking for flexibility coordination studies.
Codes available at: GitHub
I contributed to a study that quantified how energy sharing influences low-voltage distribution network operation under electrification and different electricity pricing signals, providing practical insights for network impacts and policy design.
I systematically benchmarked multiple RL algorithms for HVAC control, comparing how model availability and policy representations affect control performance and learning stability.
