A new methodology for the PMV calculation by using less input data can be a useful and interesting tool for the
indoor thermal comfort prediction. Several studies were carried out in order to evaluate and to correlate thermal
comfort to less input parameters or to outdoor conditions, but the results were not always reliable. A simplified
approach was developed in 1971 by Rohles [11], with the aim of obtaining PMV index in terms of parameters easily
sampled in the environment: air temperature and relative humidity. However, this model had an important limit: it
was only referred to sedentary activity (metabolic rate = 1.2 met) and to a fixed range of clothing thermal insulation
(Icl d0.6 clo). The Rohles model was extended to a wide range of clothing thermal insulation in [8], in which an
interesting tool was proposed for HVAC systems testing and check
Artificial Neural Networks (ANNs) could also be an interesting tool for predicting thermal comfort sensation. In
particular, the Neural Network can allow to link thermal comfort to a few monitored input data, such as outdoor and
indoor climate conditions.