How can AI and ML benefit smart buildings?
Sean Muller has worked within the technology engineering and architecture industry for 30 years. For the last 6 years, he's been involved in Enterprise Architecture, specifically around emerging technologies and innovative technologies such as Al, ML, digital twinning, automation and robotic process automation.
We interviewed Sean as part of our podcast series; you can listen here or read on for some of the top highlights from the conversation.
There is a huge amount of data collected in a building every second; unless you have an army of people reviewing all of the events, you cannot identify important findings without AI and ML, Sean states.
He explains that in a smart building, the masses of data bring with it both challenges and opportunities. "AI and ML provide tools that allow managers to deal with things such as predictive analytics; the technology can begin to predict maintenance failures and make predictions about lifetime wear of the building and its components".
Sean points out that "this technology will not replace people but will assist them and optimise their decision-making ability".
"Over the next 5-10 years, I believe AI and ML will be the ultimate assistant to help validate a decision. You'd ask your assistant (who has access to a lot more data than a single person does), and it will either reinforce the decision or provide a different viewpoint that allows the building manager to change their plan and go in a different direction".
"The beauty of machine learning is that when it's first developed and installed within a building system, it might be 80% effective; it will be generalised across a range of different buildings. Once you deploy the technology in one building, if learning is enabled, it gets better for that building and could rise to as high as 90% or even 99% accurate for making recommendations for that building. If you lock the machine learning, it can't learn from the data; all it can do is provide a prediction, based on its original training".