Neural model for air exchange in habitable rooms.

Recent years have brought the popularity of methods in which neutral networks are applied. They seem of particular importance while dealing with diagnosing, predicting and estimating. Those methods rely on collected data base, simulation and interpolation in the so-called learning process. There are attempts at neutral network application in building engineering. The paper sums up the initial phase of research on neural applications in the air exchange estimation. Due to substantial experimental database, one is able to make all the calculations whichever might be necessary, for whatever prevailing conditions, both controlled or random.