<img height="1" width="1" style="display:none;" alt="" src="https://dc.ads.linkedin.com/collect/?pid=235914&amp;fmt=gif">

Machine Learning: How Our System Reads Your Building in 7 Days

Apr 13, 2017 12:06:37 PM

From the moment it is installed, the 75F system begins collecting data from your building and transmitting it to the cloud. We collect over 600 data points every minute, along with a weather feed, so that we understand the complete picture of what is going on inside and outside the building.

Smart Stat with Weather FeedDay 1

On the first day the 75F system is installed, we don’t know anything about your site. The control algorithm will begin with standard Proportional-Integral-Derivative (PID) control logic, which is a common control algorithm used in industry. On the first night, we observe our fledgling model. We begin to see patterns in the empirical data, such as the building's thermal envelope based on the weather. Our algorithm looks for things such as the desired temperature versus actual zone temperatures, discharge air temperatures and conditioning rates.

With insight from the algorithm, we begin to observe cause and effect in the system. We then craft an adjusted control strategy using predictive analytics to use for the next day’s instructions – a process called machine learning. Machine learning automatically analyzes the data model of your building. Using algorithms that learn from big data, machine learning provides computers the tools to discover insights without requiring programming.

Day 2

Our predictive strategy is applied and by the end of the day, we can see if our adjusted control strategy worked. By now, our empirical dataset is growing and we have more to observe. Depending on the results of our control strategy, we ask questions such as, "What further proactive measures should we take to be even better?" 

System in Balance.png

Days 3-6

We are continuing to collect data and are watching carefully to be sure that the system is doing what we expect. If we need, we can make remote adjustments to assist the system.

Day 7

After seven days, we have filled up our empirical dataset and have enough information to make a comparison between our baseline starting point and the improvements the 75F system has made. We are able to make intelligent observations about the building and make improvements as necessary to ensure things are running as optimally as possible.

heatmap-margin.png

Of course, the longer the system is running, the higher the degree of confidence. The more data is collected, the smarter our system becomes. Unlike traditional thermostats which comprehend nothing more than a static set of results, a smart thermostat learns and grows with your building, becoming more efficient every day.

Learn What Our System Can Do

Sarah Baker

Written by Sarah Baker

Lists by Topic

see all