Sparsity and Non-Negativity Constrained FTIR Spectroscopic Analysis for VOC Detection

Fourier Transform Infrared (FTIR) spectroscopy faces significant challenges in detecting volatile organic compounds (VOCs) within complex environments due to cross-absorption and spectral overlap among components. To address these challenges, an enhanced nonlinear least squares algorithm (SN-NNLS) that integrates sparsity and non-negativity constraints is proposed. The sparse regularization term effectively separates gas components with overlapping absorption features, improving the accuracy of multi-component analysis. Meanwhile, the non-negativity constraint ensures physically meaningful results by eliminating negative concentration estimates, enhancing the reliability of the outcomes. Additionally, the algorithm incorporates dynamic polynomial degree adjustments and nonlinear correction techniques to handle the nonlinear characteristics of diverse spectral datasets, further enhancing its adaptability and robustness. Experimental results demonstrate that the SN-NNLS algorithm significantly improves the precision, stability, and robustness of VOC concentration measurements. This method offers a reliable and efficient solution for quantitative infrared spectral analysis in complex environments.