Application of a fiber Bragg grating temperature sensing method based on support vector regression optimized by a...

Area of Science:

• Optoelectronics
• Sensors and Instrumentation
• Machine Learning Applications

Background:

• Fiber Bragg Gratings (FBGs) are widely used for temperature sensing.
• Traditional FBG methods often require thermal equilibrium for accurate readings.
• Real-time temperature monitoring demands faster and more precise sensing techniques.

Purpose of the Study:

• To investigate a novel GA-SVR method for FBG temperature sensing.
• To enable real-time temperature retrieval before thermal equilibrium.
• To assess the method's accuracy and speed for constant and decreasing temperatures.

Main Methods:

• Simulated application of GA-SVR with FBG temperature sensing.
• Utilized transient FBG wavelength and its change rate for temperature retrieval.
• Constructed training datasets using FBG sensor temperatures higher and lower than ambient temperature.

Main Results:

• Accurate real-time sensing of constant and decreasing ambient temperatures with 0.32°C accuracy.
• Identified unique singularity points in wavelength change rate for FBG temperatures higher than ambient.
• Achieved a demodulation time of 0.002 s, significantly faster than the FBG sensor's time constant.

Conclusions:

• The GA-SVR method provides high-accuracy, fast real-time FBG temperature sensing.
• FBG sensor temperature changes are dominant in random ambient temperature variations.
• The method is crucial for future high-performance, real-time temperature monitoring applications.