ARTIFICIAL NEURAL NETWORK MODEL FOR PMV PREDICTION

Research into thermal comfort may be conducted in the laboratory or in the field. The two approaches are different and are both necessary. The scientific attraction of laboratory experiments is that it is possible to isolate one or two variables and study the effect on subjects of changing their values while keeping all other environmental variables constant.

PMV prediction model for field survey is more practical and useful. But this model is complex in mechanism and very difficult to cope with. Subjects should be assigned randomly to different conditions , and it is necessary to avoid the production of false results. Many efforts have been made in this field, but are not well satisfied on their concept clarity, calculation simplicity , process dynamic characteristics ,resultant precision, etc..

An Artificial Neural Network (Much similar to neural network of human body) for PMV prediction of Artificial Neural Network to solve all puzzles we met before. Concrete analysis and methods are provided and discussed here in order to make great improvement on traditional PMV prediction method for field survey. The principle of an integrate PMV predictor and its possible applications are also showed within this paper.