Today’s electricity consumption in residential buildings accounts for about 40% of the overall electricity consumption worldwide. Thus, reducing electricity consumption is a matter that should be everyone’s concern. A better environment with significantly reduced CO2 emissions should be the goal.
In any residential building, reducing electricity consumption is (to a great extent) a matter of appropriately scheduling the use of all heating/cooling sources. To this end, today’s research efforts are moving towards the design and deployment of automatic/smart thermostats which will optimize the involved scheduling decisions. How?
Before being able to set-up and solve such complex optimization problems, we must first be able to predict the evolution of the house temperature given a heating/cooling schedule. The formulation of such predictions may not be trivial in general. A recently published paper of mine in the European Control Conference 2014 in Strasbourg, France, discusses and compares linear black-box models models with more detailed physics-based nonlinear regression models.
When such more detailed models should be considered and what is the benefit with respect to prediction error?
Further details are given in the research article and also in the paper’s first presentation.
My Journey with Intelligent Decision Support Systems 
Leave a comment