Neural Network Application for Air Exchange

The process of air exchange can be described through both planar and spatial network system. It depends on a few random variables (those related to climate) and also on controlled variables (i.e. those like arrangement, etc.). Consequently, the air exchange problems are solved only approximately. In order to avoid that, a neural model was applied as well as estimation in the so-called learning process with simultaneous weight correction. On the basis of comparison with experimental data it can be claimed that solutions presented in the paper demonstrate high result congruence. Neural network applications in air exchange offer quick diagnosis and can be adapted to all climatic and environmental (controlled) conditions.