Photoacoustic spectroscopy method for simultaneous detection of acetylene and ammonia

Abstract This paper proposes a simultaneous detection method for acetylene (C 2 H 2 ) and ammonia (NH 3 ) based on photoacoustic spectroscopy (PAS) technology. To address the issue of cross-interference caused by the overlapping absorption lines of the gases near 1531.6 nm, a nonlinear second-harmonic theoretical model was established, revealing the nonlinear coupling characteristics of the harmonic signals in gas mixtures (e.g. the evolution of a five-peak waveform into a four-peak waveform with changing concentrations). The experiment employed a single laser (modulation frequency 4202.5 Hz) and a photoacoustic cell with a resonant frequency of 8505 Hz, combined with lock-in amplification technology to acquire the second-harmonic signals. To solve the spectral aliasing challenge, a multilayer perceptron (MLP) neural network was designed, taking 167 harmonic data points as input, and achieving dual-gas concentration inversion through end-to-end training. On a validation set of acetylene (20–100 ppm) and ammonia (20–90 ppm) gas mixtures, the average relative errors reached 2.87% and 2.44%, respectively, significantly outperforming traditional linear methods. This method eliminates the need for multiple lasers or complex photoacoustic cell structures, providing an efficient solution for simultaneous multi-gas detection.