Indoor air pollution poses a major global health threat, with complex chemical components and low detection standards being core issues. Due to intrinsic cross-sensitivity, a wide range of commonly used resistive gas sensors struggle to accurately identify multiple gases in mixtures. Simultaneously identifying gas species and their concentrations has been a significant challenge in the field of gas detection. Herein, the strategy involves creating a sensor chip and employing a pattern recognition algorithm to accurately identify the species and concentrations of gases. To address the cross-sensitivity problem of individual gas sensors, a carbon-based FET gas sensor chip (GSC) with multiple sensing gates is proposed. This chip is expected to provide sufficient gas sensing response data and accurately detect the species and concentrations in multicomponent gas mixtures. Furthermore, by combining support vector machine (SVM) algorithms with multiple linear regression (MLR), an identification model for multiple gas species and their concentrations is developed and applied to the recognition of gaseous NH3, H2S, and HCHO. Particularly, the accuracy of identifying single gas species and their concentrations exceeds 90.1%, and the accuracy of identifying target gas species and their concentrations in mixed gases exceeds 76.51%. Our work is expected to break through the bottleneck of simultaneously identifying gas species and their concentrations in multicomponent gas mixtures.
Liu et al. (Mon,) studied this question.