Towards a Deep Reinforcement Learning Approach to Instil Energy-Saving Behavioural Changes for Building Occupants in the Tropical Region
The building sector is one of the major energy consumers in Singapore. According to BCA’s report, Singapore’s building sector consumes up to 38% of the nation’s electricity, and the Heating, Ventilation and Air Conditioning (HVAC) systems account for 40~50% of the total energy consumed by buildings. As a result, the reduction of the energy expenditure of HVAC system would save a lot of energy and also lower the electricity bill. However, the HVAC system directly impact the occupants’ indoor thermal comfort. For example, in Singapore, setting higher room temperature will definitely reduce the cooling energy consumption, but it will also affect occupants’ comfort levels. It follows that it is a challenge to strike the trade-off between thermal comfort and energy saving while instilling occupants’ behavioural changes.
This research proposes to develop the online personalized behaviour influence models, via the deep reinforcement learning (DRL) technique, and apply them for intelligent HVAC system control as well as energy-saving behavioural studies to drive the green buildings in the tropical region (e.g., Singapore). Leveraging deep data analytics over information acquired from smartphone crowdsourcing, in-situ wearable measurements and human behavioural experiments, we plan to develop and validate an Intelligent Energy-Saving Persuasive (iESP) system, with the following technical aims:
- To validate the effects of existing persuasive systems in the tropical region;
- To develop the online personalized behaviour influence models via human behavioural experiments and deep reinforcement learning technique;
- To derive a novel human behaviour influence mechanism to motivate occupants’ behavioural changes to form sustainable energy-saving habits in Singapore.
Our solution builds upon our expertise in pervasive sensing, big-data analytics, artificial intelligence (AI), and focuses on applied R&D in smart buildings. First, we will develop a human-centric solution to leverage wearables (e.g., wristband) and mobile devices (e.g., smartphone) for crowdsourcing user preferences and insitu measurements. Second, we will conduct national surveys, human behavioural experiments, and AI-based deep analytics to build the online personalized behaviour influence models for energy-saving behavioural studies. Finally, analytical insights will be validated in real testbeds (e.g., Innovation Lab at NTU, Zero Energy Building at BCA). In this project, our expected deliverables include an iESP system with light-weight mobile Apps (Android & iOS). Moreover, we team up with Prof Rosenthal, from the Wee Kim Wee School of NTU, to design the data collection experiments, and aim to explore the effective ways to instil human behavioural changes in the long-term run. Our proposal, when adopted, is expected to facilitate BCA’s vision to “engage building tenants and occupants more actively to drive energy consumption behavioural changes”.
Our proposed solution stands out for its high readiness and socioeconomic impact on energy-saving management in Singapore. First, our solution addresses a fast growing (with a CAGR of 33.7%) smart building market with an estimated market size of $31.74 billion by 2022. Second, since our system only collects information from existing devices, it can easily enlist the participation of building owners and tenants, without requiring major modifications in building management infrastructures. Third, as a major urban centre in the tropics with a highly built-up environment, successful adoption of this data-driven model will put Singapore at the frontier of building energy management. The ability for the model to link the new thermal comfort parameters and behavioural change factors with facilities efficiency data may also have a direct bearing on the productivity and well-being of its workforce while curbing building energy consumption and helping Singapore’s commitment to reducing greenhouse gas emissions without penalty on its economic performance. Finally, the technology and talent pool, generated by in-house research and development efforts, will attract more international companies to Singapore, boosting economic growth in this growing sector.