Multicomponent Alkane IR Measurement System Based on Dynamic Adaptive Moving Window PLS

Infrared (IR) spectroscopy for multicomponent alkane measurement has shown excellent potentials in unconventional oil and gas exploration. However, the cross-sensitivity of alkane IR absorption and the complexity of unconventional oil and gas components place extreme demands on quantitative IR spectroscopy models. To cope with these problems, an adaptive moving window partial least square (AMW-PLS) modeling method for the quantitative analysis of IR spectra is proposed in this article. Based on the designed moving control function and dual endpoint independent moving strategy, AMW-PLS can quickly determine the optimal local modeling interval location and width by tracking the accuracy of the calibration model dynamically. By transposing the proposed AMW-PLS into the designed IR gas logging system in this article, 88 samples of two different mixture types were tested and compared with the other three typical analytical methods. The experimental results show that AMW-PLS can dynamically track the IR characteristic shifts caused by the changes in the gas mixture component complexity and concentration distribution, and thus achieves the best performance. The average relative error of AMW-PLS method for five alkanes in three standard industrial mixtures under strong interference is 3.3651%, which provides new technical support for early discovery of oil and gas reservoirs.