An E-Nose-based indoor air quality monitoring system: Prediction of combustible and toxic gas concentrations

Abstract

A system for monitoring and predicting indoor air quality level is proposed in this paper. The system comprises a computer with a monitoring program and a sensor cell, which has an array of metal oxide gas sensors along with a temperature and humidity sensor. The gas sensors in the cell have been chosen to detect only hydrogen, methane, and carbon monoxide gases. Methane was selected as a representative for indoor combustible gases, and carbon monoxide was used to represent indoor toxic gases. Hydrogen was used as an interfering (and also combustible) gas in the study. A number of experiments were conducted to train the three artificial neural networks of the monitoring system. The networks have been trained using 80% of the gathered data with the Levenberg-Marquardt algorithm. The results of this work show that the performance rate of the proposed monitoring system in determining gas type for the limited sample space is 100% even when there is an interfering gas such as hydrogen in the environment. The trained system can predict the concentration level of the methane and carbon dioxide gases with a low absolute mean percent error rate of almost 1%.