Narrow-band multi-component gas analysis based on photothermal spectroscopy and partial least squares regression method

Yang Zhou, Meng Jiang, Wei Dou, Donghui Meng, Chao Wang, Junhua Wang, Xuefeng Wang, Lichen Sun, Shoulin Jiang, Feifan Chen, Wei Jin

Research output: Journal article publicationJournal articleAcademic researchpeer-review

4 Citations (Scopus)

Abstract

In spectroscopic multi-component gas sensing, the crosstalk in the overlapping region of gas absorption spectra will affect the measurement accuracy. This paper reports a new multi-component gas analyzing method based on partial least squares regression (PLSR), which can retrieve target gas concentration and the composition and concentration of interfering gas from an overlapping narrow-band spectrum. Two overlapping absorption spectra (acetylene and ammonia) around 1530 nm are selected to simulate the absorption crosstalk phenomenon. We built a fiber-optic photothermal interferometry system to verify the method on two-component gases in the overlapping region. The training of the PLSR model was based on 63 sets of photothermal second-harmonic signals, which acquired from the gases composed by 100–700 ppm ammonia and 100–900 ppm acetylene. The prediction error of PLSR model achieved 3.84 % when tested by extra data.

Original languageEnglish
Article number133029
JournalSensors and Actuators B: Chemical
Volume377
DOIs
Publication statusPublished - 15 Feb 2023

Keywords

  • Hollow-core fibers
  • Multi-component gas analysis
  • Narrow-band
  • Partial least Squares regression
  • Photothermal spectroscopy

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Instrumentation
  • Condensed Matter Physics
  • Surfaces, Coatings and Films
  • Metals and Alloys
  • Electrical and Electronic Engineering
  • Materials Chemistry

Fingerprint

Dive into the research topics of 'Narrow-band multi-component gas analysis based on photothermal spectroscopy and partial least squares regression method'. Together they form a unique fingerprint.

Cite this