Electro-Optical Nose for Indoor Air Quality Monitoring

Abstract

Nowadays, indoor air pollution is a major problem that affects human health. For that reason, measuring indoor air quality has an increasing interest. Electronic noses are low-cost instruments (compared with reference methods) capable of measuring air components and pollutants at different concentrations. In this paper, an electro-optical nose (electronic nose that includes optical sensors) with non-dispersive infrared sensors and metal oxide semiconductor sensors is used to measure gases that affect indoor air quality. To validate the developed prototype, different gas mixtures (CH4 and CO2) with variable concentrations and humidity values are generated to confirm the discrimination capabilities of the device. Principal Component Analysis (PCA) was used for dimensionality reduction purposes to show the measurements in a plot. Partial Least Squares Regression (PLS) was also performed to calculate the predictive capabilities of the device. PCA results using all the measurements from all the sensors obtained PC1 = 47% and PC2 = 10%; results are improved using only the relevant information of the sensors obtaining PC1 = 79% and PC2 = 9%. PLS results with CH4 using only MOX sensors received an RMSE = 118.8. When using NDIR and MOX sensors, RMSE is reduced to 19.868; this tendency is also observed in CO2 (RMSE = 116.35 with MOX and RMSE = 20.548 with MOX and NDIR). The results confirm that the designed electro-optical nose can detect different gas concentrations and discriminate between different mixtures of gases; also, a better correlation and dispersion is achieved. The addition of NDIR sensors gives better results in measuring specific gases, discrimination, and concentration prediction capabilities in comparison to electronic noses with metal oxide gas sensors.