Central HVAC units have dominated home and work spaces for decades. And although they prove to be very effective at maintaining temperature, the subjective nature of comfort tends to be an issue in large workspaces with several people. For these purposes, you can opt for a Personal Comfort System or PCS, such as heated seating and fans that can help cater to every individual’s needs.
However, PCSs are limited in their effectiveness due to a communication gap between the central HVAC and individual PCS units. The constant temperature these systems provide does not consider changes within the outside environment either, and they require manual adjustment. This lack of integration can be bridged with AI and the Internet of Things. In this article, we’ll discuss proposed models for integrating PCSs with a central HVAC.
A search algorithm finds the shortest route to a problem when searching for possible solutions. For HVAC, it would find the best operational time for maximum comfort by intuitively cooling or heating the space before work hours begin. This considers people's preferences by determining a set of parameters that provide the most comfort for everyone; for example, by learning how the HVAC is adjusted throughout the day manually and incorporating those changes into an automated system.
While the search algorithm can find multiple ways to adjust the central HVAC automatically, only a handful of those answers would make intuitive sense for human comfort. Computers operate based on binary TRUE and FALSE logic, but human reasoning is much more advanced.
The AI has to make decisions based on more complex logical inferences. For example, if the outside temperature drops at night, what new temperature should the HVAC aim for, and under what circumstances should this answer differ? This will result in automation that eliminates the need for human input.
Machine learning is a sub branch of Artificial Intelligence that interprets data and creates models that best simulate a human brain.
Accurate data interpretation is essential to ensure that the AI controlling the heating or cooling of the HVAC does not overstate or understate the desired temperature of a workspace. An HVAC system must be able to make educated decisions based on the aforementioned logical inferences. By using machine learning, it is possible to train a model that would alter the parameters of the HVAC as if it were an actual human.
Due to the variation in human preference, PCSs are used to create a local environment that is made specifically for an individual. For example, if you find the air conditioning of your space to be too cold, you can use a smart foot or seat warmer to get a more desirable temperature. When you add in AI to the mixture, manual settings will contribute to the AI’s training and turn on and off automatically as they learn your preferences with time.
These PCSs need to be connected to the central HVAC system to account for the dynamic environment outside. For an algorithmic approach to fully succeed, all of these devices must be linked to one another and controlled by a singular AI for the most accurate control. Having PCSs connected to an HVAC via the internet or local LAN could be one way of achieving this.
There are several difficulties to note when implementing such a system. Such an advanced system requires significant time and memory. Secondly, PCSs alone are meant to be inexpensive devices that use a fraction of the power of an HVAC and provide individual comfort. Adding AI functionality would defeat this purpose due to higher manufacturing and operating costs.
Secondly, algorithms require data from more accurate sensors to function as intended due to the accurate control that we desire. These sensors add to the cost of a PCS or HVAC system. Buying inexpensive replacement pump parts and sensors for a traditional HVAC system is much more cost-effective. An AI-powered solution, on the other hand, would require more money for repairs.
While current research on automatic control is promising, it is yet to be seen if it can be implemented on a large scale. Still, decreasing costs of computer parts and more cost-effective manufacturing techniques may make such systems a norm in the near future.