Full-Range Ultrasensitive Fiber-Optic Hydrogen Detection via Twin Vernier Amplification and Lightweight CNN

Abstract

Reliable hydrogen monitoring requires sensors that combine high sensitivity, wide dynamic range, and resistance to environmental interference. Here, we report a monolithically integrated dual-Fabry-Pérot interferometer (dual-FPI) fiber-optic hydrogen sensor that couples twin Vernier spectral amplification with a lightweight dual-branch fusion convolutional neural network (DBF-CNN). The sensor incorporates a polydimethylsiloxane (PDMS)-filled temperature reference cavity and a PDMS/Pd-WO3 hydrogen-sensitive cavity on a single fiber, enabling intrinsic temperature compensation and humidity-insensitive detection without external reference devices. Digital twin Vernier amplification provides an approximately 10-fold sensitivity enhancement, giving a sensitivity of -62.989 nm/% and a limit of detection of 5.33 ppm. To resolve the wavelength ambiguity imposed by the free spectral range (FSR), we combine physics-guided spectral data augmentation with a lightweight DBF-CNN regression model to achieve end-to-end 0-100% volume fraction hydrogen concentration inversion, with an R2 of 0.999443 and a root mean square error (RMSE) of 0.682%. This integrated sensing strategy simultaneously addresses sensitivity, detection range, and environmental cross-sensitivity, offering a compact approach for fiber-optic hydrogen safety monitoring.