GA-HIDMS-PSO-BPNN Model-Based Suppression of Cross-Interference in Absorption Spectra for Dual-Gas Sensing

Cross-interference from overlapping absorption spectra of different gases significantly limits the precision and stability of gas concentration detection using traditional absorption spectroscopy. This study proposes a novel model integrating the global search capability of the genetic algorithm (GA), the local optimization ability of the heterogeneous improved dynamic multiswarm particle swarm optimization (HIDMS-PSO), and the nonlinear modeling of the backpropagation neural network (BPNN), termed the GA-HIDMS-PSO-BPNN model, to solve the cross-interference problem in tunable diode laser absorption spectroscopy (TDLAS) for a dual-gas sensor detecting CH₄ and CO. In a 1500 s stability test, the model demonstrates reliable performance, achieving standard deviations of 24.7753 ppm at 5000 ppm of CH₄ and 0.3075 ppm at 30 ppm of CO. These results verify the effectiveness of the model in improving the accuracy and reliability of the dual-gas sensor. It provides a robust approach for maintaining high-precision concentration measurements under cross-interference conditions and offers an efficient solution for multicomponent gas analysis.