Time-Efficient Convolutional Neural Network-Assisted Brillouin Optical Frequency Domain Analysis.

Area of Science:

• Optics and Photonics
• Artificial Intelligence
• Materials Science

Background:

• Brillouin optical frequency domain analysis (BOFDA) is a technique for sensing.
• Conventional BOFDA methods for temperature extraction can be time-consuming.
• Noise can affect the accuracy and efficiency of BOFDA systems.

Purpose of the Study:

• To develop a novel, time-efficient temperature measurement method using machine learning.
• To enhance the performance and robustness of Brillouin optical frequency domain analysis (BOFDA).
• To investigate the application of convolutional neural networks (CNNs) in signal post-processing for BOFDA.

Main Methods:

• Implementation of a convolutional neural network (CNN) for signal post-processing in BOFDA.
• Comparison of the CNN-based approach with conventional Lorentzian curve fitting.
• Evaluation of the CNN's robustness against noise in temperature measurements.

Main Results:

• The CNN-based method significantly accelerates temperature extraction in BOFDA.
• Measurement time is reduced by more than nine times compared to traditional methods.
• The CNN approach demonstrates enhanced performance and robustness, particularly in noisy conditions.

Conclusions:

• Machine learning-assisted BOFDA offers a substantial improvement in measurement speed.
• The proposed CNN method provides a more efficient and robust alternative for temperature sensing.
• This advancement paves the way for real-time monitoring applications requiring rapid temperature measurements.