Dr Chaoqun Zhuang's recent CSIC research talk, titled "Decarbonising Building Energy Systems through Stochastic Modelling & Risk-aware Control Optimisation," explored cutting-edge methods to make buildings more energy-efficient and environmentally friendly. The focus of the talk was on using digital twins, which are virtual replicas of physical systems, to better manage and optimise energy use in buildings.
Zhuang's work highlights how digital twins can monitor, simulate, and optimise building systems, helping to reduce energy consumption and operational risks. Traditional models used in digital twins often rely on complex physics-based simulations, which are resource-heavy. On the other hand, simpler data-driven models may miss important physical details.
To address this, Zhuang introduced a new approach using deep Gaussian process emulators to model individual system components, which are then interconnected based on their physical associations and varying functional complexities. This method balances both data and physical laws to better predict and manage building energy performance while also reducing the time and computational power needed for simulations.
The research demonstrates that these advanced models significantly improve energy efficiency in buildings and provide a more effective pathway toward achieving decarbonisation goals. By linking deep Gaussian processes, the approach not only enhances the accuracy of predictions but also helps mitigate operational risks, marking a crucial step in the broader effort to combat climate change and reduce carbon emissions.
Watch video here
Dr Chaoqun Zhuang is a postdoctoral research associate in Energy Efficient Cities initiative, Department of Engineering, University of Cambridge.