Improved SVR based on CARS and BAS for hydrocarbon concentration detection

Abstract

As one of the main means of discovering and evaluating oil and gas reservoirs, gas logging technology plays an irreplaceable role in oil and gas reservoir exploration. However, with the increasing complexity of oil and gas reservoir exploration, gas logging technology has gradually shifted from chromatographic gas logging to infrared spectral gas logging. Infrared spectroscopy logging technology can not only quickly analyze the content of hydrocarbon components in formation fluids, but also has the advantages of no pollution, high detection efficiency, and accurate analysis. However, there are many types of hydrocarbon gases in formation fluids and a wide range of concentration ranges. How to ensure that infrared spectroscopy can accurately measure the content of hydrocarbon gases is an urgent problem to be solved. To solve this problem, this paper proposes the beetle antennae search algorithm (BAS) combined with the competitive adaptive reweighted sampling (CARS) to improve the SVR for modeling. Firstly, the characteristic wavelength points of the sample set are obtained by CARS, and then the parameters of SVR are optimized by BAS, and then the corresponding high, medium and low concentration prediction models are established. The experimental results show that the accuracy and efficiency of the proposed CARS-BAS-SVR method in the quantitative analysis of hydrocarbon gas components are better than those of the traditional method. Its prediction accuracy for low-concentration hydrocarbon gases is greatly improved, and the overall average error is less than 1‰. It effectively improves the detection accuracy of infrared spectroscopy logging technology, and provides support for the further promotion of infrared spectroscopy in gas logging.