A comprehensive evaluation method for indoor air quality of buildings based on rough sets and a wavelet neural network

Abstract

Understanding the level of indoor air quality is very important to improve the quality of air that people breathe indoors. In this paper, a comprehensive evaluation method combining rough sets and a wavelet neural network is proposed to evaluate the indoor air quality of buildings. Through on-site inspections of the indoor air in six large shopping malls in Beijing, Wuhan and Guangzhou, raw data of the environmental parameters affecting the indoor air quality of large shopping malls are obtained. First, rough sets are used to reduce the dimension of features that affect indoor air quality by removing unimportant features, and important environmental parameters that affect indoor air quality are obtained. These important environmental parameters are used as input parameters of the wavelet neural network. Then, the structure of the wavelet neural network is determined, and an evaluation model of the indoor air quality of buildings based on rough sets and the wavelet neural network is established. Finally, the model is applied to the evaluation of indoor air quality in large shopping malls, and the back propagation neural network, fuzzy neural network and Elman neural network are introduced for comparison of the testing accuracy of the wavelet neural network in the sample testing stage. The results show that the structure of the wavelet neural network is optimized by using a rough set to reduce the redundant attributes of the data, and that the comprehensive evaluation method based on rough sets and a wavelet neural network can accurately evaluate the indoor air quality level of buildings. The results of this study have significance for and can guide the evaluation of the indoor air quality of buildings.