Application of the Principal Components Analysis to Optical Fiber Sensor for Multiparametric Detection of Strain and Torsion

Abstract

In this article, a multiparameter detection method for torsion and strain based on a single optical fiber sensor (OFS) is demonstrated combined with principal components analysis (PCA). A helical intermediate-period fiber grating (HIPFG) is used to measure the strain (0–1600 μϵ) and torsion (0–±360°) simultaneously. The wavelengthshift of the HIPFG shows a linear relationship with strain with the sensitivity of −1×10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">−3</sup> nm/μϵ, but an ambiguous relationship with torsion change. Hence, PCA is coupled to the experimental data to improve the sensing performance and the first principal component (PC1) and second principal component (PC2) are obtained. Based on PC1, the strain sensitivity and torsion sensitivity are calculated to be −0.37 μϵ <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">−1</sup> and 9050.74 (rad/mm) <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">−1</sup> . Intriguingly, multiparameter analysis for strain and torsion is obtained with the PC1 and PC2 and the simultaneous prediction of strain and torsion is succeeded. The prediction error for strain and torsion are 7.6% and 3.5%, respectively, emphasizing the link between principal components and target parameters. The method proposed provides a new entry to multiparameter sensing and analysis.