Distributed humidity fiber-optic sensor based on BOFDA using a simple machine learning approach.

Area of Science:

• Fiber optic sensing
• Optical physics
• Machine learning applications

Background:

• Relative humidity (RH) monitoring is crucial in various industrial and environmental applications.
• Traditional RH sensors face limitations in distributed sensing and cross-sensitivity.
• Optical fibers offer a robust platform for sensing applications.

Purpose of the Study:

• To demonstrate distributed relative humidity sensing in silica polyimide-coated optical fibers for the first time.
• To utilize Brillouin optical frequency domain analysis (BOFDA) for RH sensing.
• To develop a machine learning-based approach for accurate RH and temperature measurements.

Main Methods:

• Employing Brillouin optical frequency domain analysis (BOFDA) for signal acquisition.
• Utilizing linear regression, a machine learning algorithm, trained on Brillouin spectrum features (frequency shifts and linewidths).
• Applying machine learning concepts for model uncertainty estimation and feature selection.

Main Results:

• Successfully achieved distributed relative humidity sensing in polyimide-coated optical fibers.
• Demonstrated the capability of the model to simultaneously provide distributed temperature information.
• Addressed and mitigated the cross-sensitivity effects between humidity and temperature.

Conclusions:

• Distributed RH sensing in optical fibers is feasible using BOFDA and machine learning.
• The developed model offers accurate and simultaneous temperature and RH measurements.
• This technique provides a novel solution for environmental monitoring with enhanced accuracy and reduced cross-sensitivity.