A Neural Network Based Modeling of Closed Room Thermal Comfort Environmental Prediction for Sensor Hub

Abstract

Currently, environmental protection and intelligence system are topics of interest in smart homes and intelligent buildings. This work represents an environmental prediction study of the closed room which installed a split-type air-conditioning system and a newly designed sensor hub using a feed-forward back-propagation artificial neural network-based prediction for human thermal comfort and indirectly concerned to the growth or existence of diseases, contagious diseases and allergies. The RPROP algorithm was utilized to train the closed room environmental parameters that get from sensor hub especial relative humidity in the ANN model and validate the model by using cohen’s kappa and ROC curve to represents the degree of accuracy and reliability. The result obtained from the ANN trained data agreed closely with those obtained from a sensor hub with the high perfect agreement. The benefit of the ANN-based strategy in this study could be utilized for the design of a sensor hub that can realize the adaptive adjustment of environmental parameters through the air conditioning system.