The system identification of HVAC artificial neural network

Abstract

An air conditioner or AC is an apparatus that designed to adjust the temperature as well as humidity in house. A multi-functional air conditioning system which contains functions like heating, ventilation and air conditioning is referred to as “HVAC”. In this study, the purpose is to estimate the dynamic model of the HVAC system by using the Least Square (LS), Recursive Least Square (RLS) and Artificial Neural Network (ANN) techniques. The input and output data used to estimate the dynamic model in this study were obtained experimentally by previous studies. The system identification techniques were conducted based on single-input-single-output (SISO) autoregressive with exogenous (ARX) model structure. The validity of the models was investigated based on mean square error (MSE), regression and correlation tests. The results of every techniques are compared with their performance of identification the system. It is indicating that in this study, the RLS method shows the better results than LS method, however in the methods of system identification using ANN, the time-series structured the method, such as Elman Network give the best results.