Real-time Fabry-Pérot cavity length demodulation via deep learning-based compressed sensing.

Area of Science:

• Optoelectronics
• Sensor Technology
• Machine Learning

Background:

• Fiber-optic Fabry-Pérot sensors offer advantages like immunity to electromagnetic interference and high resolution.
• Accurate dynamic measurements require fast demodulation techniques, typically using white light interferometry.
• Existing methods face limitations in achieving the high sampling rates needed for advanced applications.

Purpose of the Study:

• To develop a novel, high-speed demodulation technique for fiber-optic Fabry-Pérot sensors.
• To enhance the cavity length determination accuracy for dynamic measurements.
• To enable real-time, high-resolution absolute measurements in demanding applications.

Main Methods:

• Utilized a deep learning network integrated with a compressed sensing algorithm.
• Employed a Vernier-tuned distributed feedback laser for rapid wavelength tuning across a broad spectral bandwidth.
• Achieved real-time cavity length demodulation by processing the sensor's interferometric response.

Main Results:

• Demonstrated real-time cavity length demodulation with a potential sampling rate up to 10 MHz.
• Achieved a measurement sampling rate enhancement of over two orders of magnitude compared to existing techniques.
• Validated the method's capability for high-resolution and high-speed absolute measurements.

Conclusions:

• The developed deep learning and compressed sensing approach significantly advances fiber-optic sensor demodulation.
• This technique overcomes limitations of current methods, enabling unprecedented measurement speeds.
• The technology holds substantial potential for applications demanding high-fidelity dynamic sensing.