Abstract
A new algorithm for the PMV calculation was developed using Artificial Neural Networks. Several experimental
investigations were carried out in two classrooms using both Fanger static model and adaptive approaches for the
PMV evaluation. The Artificial Neural Network was trained considering a few input parameters; specifically for the
network development only the air temperature and relative humidity were considered as experimental data. This
algorithm allows to correlate the thermal sensation to both indoor and outdoor factors and it is a useful tool for
predicting the PMV index, replacing the traditional methods with less time and cost demanding.