Measuring parameters of laser self-mixing interferometry sensor based on back propagation neural network

Area of Science:

• Optics and Photonics
• Laser Physics
• Non-Destructive Testing

Background:

• Self-mixing interferometry (SMI) is a key non-destructive sensing technique.
• Accurate determination of laser parameters, optical feedback factor (C) and line-width enhancement factor (α), is crucial for SMI performance.
• Existing methods for estimating C and α are often limited to specific feedback regimes (weak or moderate).

Purpose of the Study:

• To develop a novel method for estimating laser parameters C and α in SMI systems.
• To enable accurate parameter estimation across all feedback regimes.
• To enhance the practical applicability of SMI technology.

Main Methods:

• Implementation of a back-propagation neural network (BPNN) model.
• Training and validation of the BPNN model using simulation and experimental data.
• Testing the model's performance across various feedback regimes.

Main Results:

• The proposed BPNN method accurately estimates C and α with average errors of 2.76% and 2.99%, respectively.
• The method demonstrates robustness against noise.
• Successful estimation across all feedback regimes, overcoming limitations of previous techniques.

Conclusions:

• The developed BPNN-based method offers a universal solution for estimating C and α in SMI.
• This approach significantly enhances the reliability and accuracy of SMI sensing.
• The findings facilitate broader adoption of SMI in practical engineering applications.