Simultaneous Temperature and Relative Humidity Measurement Using Machine Learning in Rayleigh-Based Optical Frequency...

Area of Science:

• Optical sensing technologies
• Machine learning applications in metrology
• Environmental monitoring instrumentation

Background:

• Accurate simultaneous measurement of temperature and relative humidity (RH) is crucial for various applications.
• Traditional methods often require complex post-processing or separate sensors.
• Rayleigh-based Optical Frequency Domain Reflectometry (OFDR) offers a promising platform for distributed sensing.

Purpose of the Study:

• To develop and validate a machine learning (ML) model for simultaneous temperature and RH sensing using OFDR.
• To investigate the performance of a dual-segment fiber optic sensor (bare and polyimide-coated) for differential environmental sensitivity.
• To eliminate manual data post-processing and reduce analysis time in OFDR-based measurements.

Main Methods:

• Utilized a sensor unit comprising bare and polyimide-coated fiber segments with distinct temperature sensitivities.
• Developed a machine learning (ML) model to process OFDR data for simultaneous temperature and RH extraction.
• Evaluated the impact of sensor unit length and data point count on measurement accuracy (RMSE).

Main Results:

• Achieved high accuracy for simultaneous temperature and RH measurements with Root Mean Square Error (RMSE) of 0.36 °C and 1.73% RH, respectively, for a 3 cm sensor length.
• Demonstrated the effectiveness of the ML model in distinguishing and quantifying temperature and RH contributions.
• Quantified the influence of sensor configuration parameters on overall measurement precision.

Conclusions:

• The proposed ML-driven OFDR system enables accurate, simultaneous, and automated measurement of temperature and RH.
• This approach significantly reduces data analysis time compared to traditional post-processing methods.
• The developed technique offers a robust solution for environmental sensing in diverse fields requiring precise monitoring.