HVAC PIPE/DUCT SIZING USING ARTIFICIAL NEURAL NETWORKS

摘要

The main objective of this study is to demonstrate that artificial neural networks (ANN'S) serve as useful aids to Heating, Ventilating and Air-Conditioning (HVAC) system design. In the present work, the design process for sizing fluid systems in HVAC is simulated by using ANN's. Four ANN's have been constructed in a personal computer, one for air duct sizing and three for pipe sizing. The air duct network was trained to output the friction rate and duct size. The three pipe sizing neural networks product pressure drops and pipe diameters. By using the trained artificial neural networks, data can be obtained instantly with errors less than 3%. Thus, ANN's have been shown to simplify traditional methods and procedures in HVAC pipe and air duct sizing.