Research
The researchers at ADSC, in collaboration with Korea Electronics Technology Institute (KETI) have created a living lab in ADSC’s Singapore offices to track how efficiently typical office spaces use electricity. Their goal is to develop a method of mining and quantifying potential energy cutbacks in buildings, by utilizing extensive sensor information.
KETI provided funding for the project, as well as a wireless sensor hardware platform, while ADSC is designing and implementing the algorithms and software to process and analyze the sensor data.
They are using a variety of sensors (current / voltage / PIR / light / air pressure / temperature / humidity / CO2 / microphone / camera / radio spectrum) across the ADSC premises to monitor the office space, people, appliances, and their energy consumptions. The data gathered by these sensors are centrally analysed and processed. Users of the system, which include ADSC staffs as well as visitors, are able to access the data live (e.g., how much power am I consuming now?) as well as for trending analysis and future planning purposes (e.g., how the room temperature varies over time and space).
Ninety power meters have been installed in the office to measure individual plug load power consumption. In addition, five multi-purpose ambient sensors were installed to measure temperature, humidity and luminous intensity, as well as five carbon dioxide sensors, nine passive infrared (PIR) sensors for motion detection and 38 current transducers (CT) to measure electricity consumption of plug load and light load branches in the main switch board.
In the meantime, we will be installing cameras at multiple sites of the office including pantry, main and side doors, corridors and printing room. We develop computer-vision algorithms to detect (localize) human users, track people and recognize user activities based on the video streams including: enter and exit door, get water from water machine, make tea/coffee, get/put food from/into refrigerator etc.
Applications that are enabled by this living lab include:
- Personalized power-consumption assistant, with features such as detailed energy consumption breakdown and automatic PC monitors switch-off. The goal is to engage the real users.
- Fine-grained public area monitoring and control, with features such as automatic adjustment of air pump valves, lighting, or other appliances’ status based on recognized user activity.
- User-friendly demand response decisions (e.g. shifting the air conditioner / refrigerator / water cooler’s load to avoid peak hours without the users even noticing that!).
Our Approaches
In the process of setting up this living lab, we have developed enabling technologies to achieve more efficient and accurate electricity consumption monitoring, auditing, and learning. These include:
- Group/ Individual/ Activity level energy consumption breakdown.
- Quantifying achievable energy saving by utilizing occupancy and environment information from sensors.
- Learning and identifying important electrical events (e.g. Turning ON/OFF Microwave) and real-time tracking of usage information of appliances of interest.
- Extracting high-level information of people’s action or contextual information based on raw power consumption data. A dual problem (for privacy protection) is to develop methods that can process the raw consumption data so as to (selectively) hide some high-level information.
We are also investigating technologies that can make occupancy sensing more efficient and accurate, for example, by leveraging measurement results from different sensors, or through fine-grained analysis of RF signals.