Abstract
This letter presents a high-capacity demodulation system for large-scale fiber Bragg grating (FBG) sensor networks, overcoming the limitations of conventional arrayed waveguide grating (AWG)-based architectures through a hybrid wavelength-division/time-division multiplexing scheme. By expanding the AWG to 32 output channels across the C-band and incorporating a 1×8 MEMS optical switch for time‑division multiplexing, the system achieves pseudo-synchronous demodulation of 128 FBG sensors across eight channels at a rate of 250 Hz, with integration enhanced via gold-wire bonding and direct coupling. A convolutional neural network (CNN)-based demodulation algorithm was used to accurately map the 32-channel optical intensity inputs to FBG center wavelengths, achieving high-precision regression. Experimental results demonstrate a wavelength demodulation stability of approximately 0.4 pm and an accuracy of 1.2 pm. The system exhibits a balanced performance in capacity, speed, and accuracy, making it suitable for demanding large-scale sensing applications in structural health monitoring and aerospace engineering.
Funding Information
  • Natural Science Foundation of Beijing Municipality (BJXZ2021-012-00046)

This publication has 13 references indexed in Scilit: