Multipoint and Multi-physical Quantity Sensing by Multimode Fiber with Demodulation of Deep Learning

A multimode fiber (MMF) speckle sensor based on deep learning is demonstrated for sensing multi-physical quantity at different position. The mode fields at the end of MMF compose speckle patterns, which are sensitive to the various types of disturbing with different manners. Further, the MMF also carry numerous information, which make it possible sensing different physical quantity at different points in just one MMF. Here, it is demonstrated to sense stretching, bending, and temperature in a MMF of 0.5m by the deep learning methods. Its stretching precision is up to 0.005mm with the U-Net.