An automatic baseline correction method based on reweighted penalised least squares method for non-sensitive areas

Abstract

Baseline drift is a key problem in spectral analysis. The resolution of the baseline drift problem sets the stage for future spectral analyses. The baseline correction algorithm previously proposed by scholars may result in inaccurate correction in the presence of spectral overlap peak areas and noise interference, and they all have one or more parameters that need to be optimized by the user. Therefore, an Automatic baseline correction algorithm based on reweighted penalised least squares method for non-sensitive areas(NasPLS) was proposed. First, according to the High solution Transmission Molecular Absorption Database (HITRAN) and actual measurements of hydrocarbon gases, several non-sensitive areas appeared between 400 and 4000 wavenumbers. Then, based on the root mean square error(RMSE) values of the original baseline and the fitted baseline in the non-sensitive areas, the minimum RMSE is found to adapt to updating the smoothing parameters and select the optimal $\lambda$. Finally, the proposed NasPLS method was validated using simulated data and actual measured spectra of methane and ethane, the NasPLS achieves precise baseline correction in the presence of noise interference in simulated data. The results of experimental measurements show that the proposed method can accurately perform baseline corrections in complex situations.