Multi-regressor location method based on SOA ring cavity reservoir computing for distributed optical fiber sensing.

Area of Science:

• Photonics
• Optical Sensing
• Machine Learning

Background:

• Distributed optical fiber sensing systems are crucial for real-time monitoring.
• Existing methods like time-delay reservoir computing (RC) face challenges in location accuracy and training complexity.
• Semiconductor optical amplifier (SOA) based RC offers potential for enhanced performance.

Purpose of the Study:

• To propose a novel multi-regressor location method using SOA ring cavity RC for distributed optical fiber sensing.
• To improve disturbance location accuracy and reduce training requirements compared to existing techniques.
• To enable precise localization without requiring sensing data from all fiber positions.

Main Methods:

• Implementation of a time-delay RC system with an SOA ring cavity.
• Integration of multiple linear regressors in the RC output layer for global and segmented location.
• Training regressors using ridge regression across different sensing ranges.
• Disturbance location achieved by thresholding segment results and selecting the closest to the global result.

Main Results:

• The proposed method successfully enabled location determination at any position along the fiber.
• Demonstrated superior location accuracy compared to the conventional time difference method.
• Significantly reduced training requirements compared to classification-based RC methods.

Conclusions:

• The multi-regressor SOA ring cavity RC method is effective for accurate distributed optical fiber sensing.
• This approach offers advantages in location precision and training efficiency.
• It provides a robust solution for disturbance detection in optical fiber systems.