Neural network and polynomial approximated thermal comfort models for HVAC systems

► We calculated two different model approximations in order to model the PMV index. ► One approximation is based on neural networks and other is calculated through a polynomial. ► We tested both approximations with several real data-sets in order to check their efficiency. ► Both approximations are able to reproduce the PMV index dynamics. ► These approximations can be used in an easily way by controllers to maintain users' thermal comfort.