March 28, 2016 | Written by: mlcullen
A cognitive building is not just a passive shelter, but an active monitor of the human activity within them.
Buildings are the number one contributor to CO2 emissions. Worldwide, buildings consume 42% of all electricity, up to 50% of which is wasted. In addition to their impact on the environment, buildings are complex and expensive to operate. Operation costs account for 71% of the total cost of a building’s ownership, with energy costs alone representing about 30% of an office building’s total operating costs.
While our phones put the world at our fingertips, and our cars are on the verge driving themselves, our buildings lag greatly behind. “Most [buildings] are still stuck in the Stone Age,” says IBM Research scientist Joern Ploennigs, who works at IBM Research – Ireland. “We spend 90 percent of our time in buildings, but often they’re not more intelligent than a cave. We still go home and turn up the heat manually.”
Ploennigs and his team are working on smart buildings that learn from past experience to make intelligent and timely adjustments that maximize comfort and productivity while minimizing energy usage.
Smart Buildings That Learn
Cognitive buildings are not just passive shelters but active monitors of the human activity within them. They compile data to learn how and at what times people use the building, their comfort preferences and ultimately how this influences energy consumption. Based on this knowledge the building can guide occupants to a meeting room, adjust settings to optimize individual comfort and control relevant systems at peak efficiency, all at the same time.
There are obvious advantages. Take schools, where studies have shown that student attention diminishes when air quality does, and that optimal ventilation is essential for students. “Our system learns to estimate the number of students in a room,” says Ploennigs. “Now we can find optimal times to activate the ventilation before any negative effects arise because of air quality. At the same time, we waste no energy ventilating a room that is empty.”
How do these cognitive buildings learn so much about us? Through a multitude of sensors. The good news is that many modern buildings are already equipped with hundreds of devices and the Internet of Things is adding more. The bad news: these sensors are often badly labelled and provide noisy data. Most buildings are like Towers of Babel when it comes to what their myriad sensors say. “The learning in a cognitive system starts with identifying and integrating useful sensors automatically,” Ploennigs says. His team built a tool that automates the process of corralling the sea of disjointed sensors into a usable array of data using artificial intelligence.
The Power of Data
A cognitive building is a self-aware building.
This can be particularly pivotal when something goes wrong, like if an air-conditioning unit gets stuck on or lights are left burning long after everyone has left for the day. In a standard building such anomalies might be caught only when someone calls to complain about it being too cold. But in a cognitive building, an e2-diagnoser learns what normal energy consumption looks like and then compares the real energy consumption and can quickly detect abnormalities; the e2-diagnoser is able to tell the source of anomalous consumption so that it can be quickly resolved.
The real power of cognitive buildings multiplies when they do: when buildings work together on campuses, or in communities or even countries. A buildings amass and analyze more data about their operations and energy consumption, more and better benchmarks are created that can be used to identify inefficient buildings. Cognitive buildings can present their power consumption predictions to energy providers with an eye toward cost-effective power purchases and preventing peak consumption system overloads.
You can read more about cognitive and IBM Research – Ireland here.