Application of a back propagation neural network model based on genetic algorithm to in situ analysis of marine sediment cores by X-ray fluorescence core scanner

Appl Radiat Isot. 2022 Jun:184:110191. doi: 10.1016/j.apradiso.2022.110191. Epub 2022 Mar 10.

Abstract

The use of core scanners to perform X-ray fluorescence (XRF) spectroscopic analysis can only obtain the intensities of the target elements, which is not conducive for application in marine geology research. In this paper, using a core scanner, in situ measurements were performed on 15 components: Al2O3, SiO2, K2O, CaO, TiO2, MnO, Fe2O3, V, Cr, Zn, Rb, Sr, Y, Zr, and Ba. We explored the feasibility of reducing the effect of interstitial water by normalizing the original intensity using the intensity of Ca and the ratio of coherent to incoherent and attempted to introduce a genetic algorithm-back propagation neural network model and use its nonlinear fitting capability to correct the matrix effect. The prediction precision of this method was 0.6-15.4%. The proposed method is suitable for the rapid analysis of major and minor components in marine sediment core samples, while taking full advantage of the high-resolution of the XRF core scanner.

Keywords: In situ analysis; Major and minor components; Marine sediment core; Matrix effect; Neural network; X-ray fluorescence core scanner.

MeSH terms

  • Algorithms
  • Geologic Sediments* / analysis
  • Neural Networks, Computer
  • Silicon Dioxide*
  • X-Rays

Substances

  • Silicon Dioxide