Ultratrace CO/SF₆ Detection System Based on Differential Fourier Transform Infrared Spectroscopy Combined with the...

Area of Science:

• Spectroscopy
• Chemical analysis
• Materials science

Background:

• Carbon monoxide (CO) is a critical indicator gas for fault diagnosis in gas-insulated switchgear (GIS).
• Fourier transform infrared (FTIR) spectroscopy is suitable for CO detection due to its strong mid-infrared absorption, but SF₆ interference poses a significant challenge.
• Accurate detection of ultratrace CO levels in SF₆ is crucial for GIS reliability.

Purpose of the Study:

• To develop a high-sensitivity online detection system for ultratrace carbon monoxide (CO) in sulfur hexafluoride (SF₆).
• To overcome the challenge of strong SF₆ absorption interference in FTIR spectroscopy for CO detection.
• To establish a robust method for accurate CO concentration measurement in GIS environments.

Main Methods:

• Integration of differential Fourier transform infrared (DFTIR) spectroscopy for baseline correction of CO absorption signals.
• Development of a weighted sine spectral reconstruction convolutional neural network (WSSR-CNN) for feature extraction and denoising of weak spectral signals.
• Application of CNN models for CO concentration detection based on processed spectral data.

Main Results:

• The WSSR-CNN method demonstrated superior performance compared to four other models.
• Evaluation index R² reached 0.99982 (low concentration) and 0.99999 (high concentration).
• Mean absolute percentage errors were 0.97% (low concentration) and 0.22% (high concentration), with a detection limit of 13 ppb.

Conclusions:

• The developed DFTIR and WSSR-CNN system provides a highly sensitive and accurate method for detecting ultratrace CO in SF₆.
• This approach significantly mitigates interference from SF₆ absorption, improving CO detection reliability.
• The system represents a breakthrough in FTIR-based CO/SF₆ detection, offering the best reported results to date.