Machine‐Learning‐Enhanced Dual‐Diameter Sapphire Fiber Bragg Gratings for Simultaneous High‐Temperature and Strain Sensing

ABSTRACT Fiber‐optic multiparameter sensors are versatile tools for in situ health monitoring of critical components in hypersonic vehicles, including thermal protection systems and turbine engine blades. Sapphire fiber Bragg gratings (SFBGs), owing to excellent high‐temperature performance, offer a competitive solution for such applications. However, these devices suffer from cross‐sensitivity, complicating the discrimination between temperature and strain responses. Here, we propose a concise and robust sensing scheme based on dual‐diameter SFBGs (DD‐SFBGs), with machine learning (ML)‐assisted demodulation. Two line‐by‐line SFBGs are inscribed by a femtosecond laser into an etched dual‐diameter sapphire fiber. By tailoring the fiber geometry, this design overcomes the cross‐sensitivity limitation within a single sapphire fiber. The DD‐SFBGs exhibit markedly different strain sensitivities of 4.16 and 0.92 pm/με at 1450°C, which can be further tailored by etching fibers with different geometries, far exceeding the capability of silica‐based sensors. Furthermore, we employ the XGBoost gradient boosting algorithm to demodulate the multimode, wide‐bandwidth reflection spectra. This ML‐enhanced approach achieves high sensing performance, with root mean square errors (RMSE) of 0.41°C for temperature and 5.76 με for strain. This compact, reliable, and high‐precision approach thus represents a significant advance in simultaneous temperature‐strain sensing, with promising applications in aerospace structural health monitoring.