Key points are not available for this paper at this time.
The concurrent attainment of high sensitivity and a large dynamic range in measuring parameters of interest holds paramount importance in sensor development. In this paper, we demonstrate a displacement sensor based on a simple balloon-shaped optical fiber device that boasts both high sensitivity and a broad dynamic range, empowered by machine learning (ML) analysis. This ML-driven approach capitalizes on global features extracted from the measured raw spectrum, overcoming the constrained dynamic range limitations posed by the traditional dip tracking method. A root mean squared error as low as 0.193 μm is obtained within a displacement calibration range of 0∼300 μm, despite utilizing a modest training dataset. The reliable prediction of unseen displacement within the calibration range is also demonstrated. This intelligent sensor system, combining a straightforward balloon-shaped fiber device with ML analysis, holds promise for wide-ranging, high-precision displacement sensing applications across diverse fields.
Naku et al. (Tue,) studied this question.