Research

Overview

BIOR LOGO

Our Building Informatics and Operations Research (BIOR) lab aims at developing sustainable and scalable technologies and computational tools to make building and urban energy systems low-carbon, energy-efficient, energy-flexible, climate-resilient, and equitable. Our interdisciplinary research is at the interface of Building Environment and Energy Engineering (i.e., Architectural Engineering), Computer Science, and Control Engineering. We employ a multifaceted approach that encompasses data analytics & machine learning, physics-based modeling & simulation, optimization & model-based optimal controls, as well as experiments. These approaches have been deployed across a spectrum of scales, spanning from equipment- through building- and community- to city-scale.

Currently, our lab is focusing on:

  • Low-carbon building technologies for energy efficiency, flexibility, and climate resilience
  • Grid-interactive smart buildings and communities
  • Modelling, estimation, optimization, and control of building, district, and urban energy systems
  • Modelling, estimation, optimization, and control of HVAC&R systems
  • Data analytics and machine learning for buildings
  • Uncertainty analysis for buildings
  • Occupant behavior in buildings

Research Methods

Data Analytics

Data Cleaning | Data Visualization | Data Clustering | EDA

Data-driven Modelling

Grey-/Black-box | Machine Learning | Deep Learning

Physics-based Modelling

Building Thermal Dynamics | Refrigeration/HVAC systems | ODEs and State Space Representations

Computational Optimisation

Mathmatical and Metaheuristic (GA/PSO) Algorithms

Model Predictive Control (MPC)

Moving from predictive to prescriptive analytics | MPC = Data-driven Modeling + Receding-horizon Numerical Optimization + Online Feedback Control

Uncertainty Analysis

Monte Carlo technique |Stochastic Occupancy and Occupant Behavior (Markov Chain)

MPC of Building Energy Systems

For energy efficiency/energy flexibility

More info

How MPC works?

Receding Horizon Optimisation: Opt.→Decision→Measurement→Opt.→Decision→Measurement→…

More info

Stochastic Data-driven Occupancy Model

Using Heterogeneous Markov-chain Monte Carlo method

More info

Data-driven Resistance-Capacitance Building Thermal Model

Simple and ready for control/large-scale purpose

More info

Previous Next

Selected Projects

COOLER at Stanford

To enable low-carbon, electrified community energy systems using Stanford campus as a real-life testbed.

AMiDine in the UK

To devise analytics that link point measurement to whole system to address the increasingly problematic management of electrical load on distribution networks as the UK transitions to a low carbon energy system.

Quantifying Uncertainty in Aggregate Energy Flexibility of Building Clusters

To quantify the uncertainty in the aggregate energy flexibility of residential building clusters using a data-driven stochastic occupancy model that can capture the stochasticity of occupancy patterns.

MPC of Variable-speed ACs for Demand Response in Smart Grids

To directly control the operating frequency of ACs in response to high-granularity electricity price signals, i.e., 5-minute real-time electricity prices,in smart grids using MPC method.

MPC of Floor Heating Systems in Denmark

To develop model predictive control for floor heating systems to provide energy flexibiltiy. The proposed MPC can simultaneously consider all the influential variables including weather conditions, occupancy and dynamic electricity prices.