Researchers have created an AI (artificial intelligence) that conserves energy by making better use of weather forecasts. It turns smart buildings into super-smart buildings.
If the weather forecast predicts rain, you may pack an umbrella. You may bring a scarf and gloves if it forecasts a steep drop in temperature. Buildings with smart heating and cooling systems do the same, i.e., they adjust themselves according to weather forecasts.
When forecasts are not 100% accurate, buildings may end up wasting a lot of energy. We sometimes do the same, either overdressing or finding ourselves cold and wet.
Machine learning model that conserves energy
Fengqi You, a Professor of Energy Systems Engineering at Cornell University, developed a new approach. It predicts the accuracy of weather forecasts using a machine learning model. The model has a year’s worth of data on actual weather conditions and forecasts.
Prof. You combined that forecast data with a mathematical model that considers a building’s size, construction, and shape of rooms. It also considers the position of the building’s windows and the location of its sensors.
Prof. You says that the smart control system can reduce energy usage by up to ten percent. Prof. You and Chao Shang, Assistant Professor of Automation at Tsinghua University in China, wrote about their work in the Journal of Process Control (citation below).
Prof. Shang was once a Cornell postdoctoral associate in Prof. You’s lab.
The researchers conducted their study on Toboggan Lodge, a building on Cornell’s campus that is almost 90 years old.
A team of students studying for their Master’s degree helped develop the case study.
The ‘smarter’ the building the better it conserves energy
Prof. You said:
“If the building itself could be ‘smart’ enough to know the weather conditions, or at least somehow understand a little bit more about the weather forecasting information, it could make better adjustments to automatically control its heating and cooling systems to save energy and make occupants more comfortable.”
“For instance, if I know the sun is going to come up very soon, it’s going to be warm, then I probably don’t need to heat the house so much. If I know a storm is coming tonight, then I try to heat up a little bit so I can maintain a comfortable level.”
“We try to make the energy system smart, so it can predict a little bit of the future and make the optimal decisions.”
With the appropriate data, the model could detect uncertainty. It detected uncertainty not just in temperature but also in sunlight, precipitation, and differences in conditions by location.
The model conserves energy by adjusting itself according to the forecasts’ levels of uncertainty.
Weather forecasts not 100% accurate
Prof. You said:
“Even the best weather forecasting system is not going to give you the most accurate information. Plus, the weather forecast information is usually for a certain region but not a specific location.”
The system conserves energy better if you combine the mathematical programming methods and machine learning algorithms. The combination creates a control system that is more accurate – ‘smarter’ – than either is on its own, Prof. You explained.
Their framework has potential applications in irrigation control in agriculture and in building control systems, the authors say. It also has potential in indoor environmental controls in vertical farms and plant factories.
Prof. You said:
“We don’t have a perfect way to forecast the weather, so the best thing we can do is combine AI and mechanistic modeling together. These two parts have never before been harmonized in a systematic way for automatic control and energy management.”
Citation
“A data-driven robust optimization approach to scenario-based stochastic model predictive control,” Chao Shang and Fengqi You. Journal of Process Control, Volume 75, March 2019, Pages 24-39. DOI: https://doi.org/10.1016/j.jprocont.2018.12.013.
