Interference fringe suppression in tunable diode laser absorption spectroscopy based on CEEMDAN-WTD

Abstract

Owing to interference fringes in the multireflective gas cell, the detection sensitivity of a system in tunable diode laser absorption spectroscopy (TDLAS) will decrease significantly. In this work, a combined scheme of complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and wavelet threshold denoising (WTD) is proposed. Theoretical simulations were performed to validate the effects of the proposed algorithm, which was also verified via a CO2 detection experiment. After CEEMDAN-WTD processing, the noisy intrinsic mode function (IMF), pure IMF, and residual components of the detection signal were identified and reconstructed successfully. Based on analysis of the simulations, CEEMDAN-WTD algorithm improved the signal-to-noise ratio by 1.87 times and decreased the root mean-squared error by 37.6% than the moving average algorithm. For the CO2 detection system, R2 = 0.999 was determined by the calibration experiment. Additionally, based on Allan variance analysis and a long-time experiment, the limit of detection was estimated to be 3.08 ppm for an average time of 148 s and measurement accuracy of 0.65%, respectively. The obtained results sufficiently validate that the CEEMDAN-WTD algorithm can effectively suppress interference fringe noise in TDLAS.