High-efficiency FBG array sensor interrogation system via a neural network working with sparse data.

Area of Science:

• Optical Engineering
• Sensor Technology
• Artificial Intelligence

Background:

• Fiber Bragg Grating (FBG) array sensors are crucial for multi-point structural monitoring due to their multiplexing capabilities.
• Existing demodulation systems can be costly and data-intensive.

Purpose of the Study:

• To develop a cost-effective and efficient demodulation system for FBG array sensors.
• To leverage Neural Network (NN) technology for accurate wavelength interrogation.
• To address data size limitations in data-driven sensor analysis.

Main Methods:

• Encoding stress variations using an Array Waveguide Grating (AWG) into transmitted intensities.
• Utilizing an end-to-end Neural Network (NN) model for nonlinear relationship mapping.
• Implementing a low-cost data augmentation strategy to enhance NN performance with limited data.

Main Results:

• The proposed NN-based system achieves absolute interrogation of peak wavelengths.
• Superior performance is demonstrated even with small-scale datasets.
• The system effectively decodes stress variations from transmitted intensities.

Conclusions:

• The developed demodulation system offers an efficient and reliable solution for FBG array sensors.
• This approach provides a cost-effective method for multi-point monitoring of large structures.
• The integration of NN and data augmentation overcomes common challenges in sensor data analysis.