Monitoring and Regulation of Micro-Displacement Deviation in Few-Mode Beam Alignment through Mode Decomposition

TL;DR

This paper presents a simple, accurate method combining mode decomposition and machine learning for 3D fiber displacement measurement, significantly improving optical beam alignment precision with minimal computational effort.

Contribution

It introduces a novel approach integrating mode decomposition with machine learning for precise 3D fiber displacement measurement in few-mode configurations.

Findings

Coefficient of determination of 0.99 for transverse offsets

RMSE of 0.135 μm, 0.128 μm, and 2.42 μm in x, y, z directions

Single displacement calculation time of 0.0004037 seconds

Abstract

Beam alignment enables efficient, stable transmission and control of optical energy and information, which critically depend on precise monitoring and regulation of the three-dimensional (3D) relative positioning between fibers. This study introduces an approach to achieve more accurate 3D measurement of the spatial displacement between two optical fibers in a few-mode configuration, by integrating mode decomposition with a straightforward machine learning algorithm. This method leverages inherent information from the optical field, enabling precise beam alignment with a simple structure and minimal computational effort. In the 3D measurement experiment, the proposed method achieves a coefficient of determination of 0.99 for transverse offsets in the x- and y-directions, and 0.98 for air gap in the z-direction. The RMSE in x-direction, y-direction and z-direction is respectively 0.135 {\mu}m, 0.128 {\mu}m and 2.42 {\mu}m. The time for a single 3D displacement calculation is 4.037e-4 seconds. Furthermore, it facilitates single-step displacement regulation with a deviation tolerance within 0.15 {\mu}m and modal content regulation with an accuracy of 4.67%. These results establish a theoretical framework for addressing key challenges in optical path alignment, crosstalk compensation, precision instrument manufacturing, and fiber optic sensing.