Cost-effective and lightweight FBG demodulation system based on edge AI

Fiber Bragg grating (FBG) sensors are extensively employed for structural health and condition monitoring. However, the high cost and large size of conventional FBG demodulation methods have hindered the widespread adoption of FBG-based sensing technologies. To overcome these challenges, this study proposes what we believe to be a novel demodulation system leveraging a tunable laser. The system utilizes a gated recurrent unit (GRU) peak-finding algorithm, implemented on an STM32 microcontroller with edge AI capabilities, all integrated onto a compact 100×100×10 mm circuit board. This approach enables demodulation across the entire C-band, achieving a monitoring range of 70 km, a demodulation frequency of 100 Hz, and a mean absolute error (MAE) of 9.6 pm.