Discrimination of Single and Binary Gases Using an Intelligent Electronic Nose System Based on Metal Oxide Gas Sensors

The pollutants emitted from industrial facilities seriously affect the atmospheric environment and human health. Accurately detecting low-concentration gas pollutants by single metal oxide gas sensors remains a significant challenge. In this study, we developed a novel electronic nose system to precisely discriminate components and concentrations of single and binary gas mixtures. The specific sensor array was constructed with six sensors based on SnO 2 and Co 3 O 4 sensing materials. The dataset collected from the electronic nose system was analyzed using k-nearest neighbor (KNN), support vector machine (SVM), random forest (RF), and logistic regression (LR) algorithms. For low-concentration ethanol, ammonia, and toluene, the KNN model accurately discriminated gas components and concentrations, achieving a five-fold cross-validation accuracy of 97.2%. As the number of sensors increased from 3 to 6, the accuracy for the types and concentrations of single and binary mixtures rose from 78.9% to 97.2%. The discrimination results for low-concentration gaseous pollutants demonstrate great potential for real atmospheric quality monitoring and provide an effective solution to the precise detection by metal oxide sensors. Based on SnO 2 /Co 3 O 4 six-sensor array, a novel electronic nose empowered by the KNN algorithm has achieved an accuracy rate of 97.2% in in discriminating low-concentration ethanol/NH 3 /toluene mixtures, which is conducive to achieving precise monitoring of air pollutants. • Fabricated sensors were highly sensitive and selective to particular gases. • The sensor array could distinguish different gases and mixtures. • KNN model achieved 100% accuracy for single gases. • KNN model achieved 97.2% accuracy for mixed gases.