Smart Gas Sensors: Materials, Technologies, Practical ‎Applications, and Use of Machine Learning – A Review

The electronic nose, popularly known as the E-nose, that combines gas sensor arrays (GSAs) with machine learning has gained a strong foothold in gas sensing technology. The E-nose designed to mimic the human olfactory system, is used for the detection and identification of various volatile compounds. The GSAs develop a unique signal fingerprint for each volatile compound to enable pattern recognition using machine learning algorithms. The inexpensive, portable and non-invasive characteristics of the E-nose system have rendered it indispensable within the gas-sensing arena. As a result, E-noses have been widely employed in several applications in the areas of the food industry, health management, disease diagnosis, water and air quality control, and toxic gas leakage detection. This paper reviews the various sensor fabrication technologies of GSAs and highlights the main operational framework of the E-nose system. The paper details vital signal pre-processing techniques of feature extraction, feature selection, in addition to machine learning algorithms such as SVM, kNN, ANN, and Random Forests for determining the type of gas and estimating its concentration in a competitive environment. The paper further explores the potential applications of E-noses for diagnosing diseases, monitoring air quality, assessing the quality of food samples and estimating concentrations of volatile organic compounds (VOCs) in air and in food samples. The review concludes with some challenges faced by E-nose, alternative ways to tackle them and proposes some recommendations as potential future work for further development and design enhancement of E-noses.

Keywords:

Gas sensor arrays , E-nose , disease diagnosis , leakage detection , machine learning , Volatile organic compounds‎