Infrared spectral denoising of gas logging based on median filtering and wavelet threshold denoising

With the development of optoelectronic information technology, more and more wellsites are using infrared spectroscopy to detect gases in drilling fluids, transforming towards equipment automation and oilfield digitization. However, the unavoidable noise will affect the quality of the spectral data of the detected gas, which will interfere with the qualitative and quantitative analysis of the gas. In view of the fact that a single denoising method can not achieve better denoising accuracy and denoising effect, a hybrid denoising method based on median filtering and improved wavelet threshold denoising is proposed in this paper. Firstly, the denoising ability of different denoising transmitters is preliminarily verified by artificially adding noise to the simulated data. The results show that the hybrid denoising method proposed in this paper is the best, and the signal-to-noise ratio (SNR) is 8.9183 and the root mean square error (RMSE) is 8.9183. Then, using the measured data of methane, different filtering methods combined with PLS quantitative analysis model are used to predict the high, medium and low concentration of methane. The results show that the denoising effect of the proposed method is better than that of other denoising methods at different concentrations. The predicted results R-Squared (R²) of high concentration, medium concentration and low concentration were 0.9455, 0.9946 and 0.9939 respectively. It improves the quality of spectral data, which can further enhance the robustness of the automated system and is important for improving the efficiency of intelligent decision-making.