Comparison of regression methods for transverse load sensor based on optical fiber long-period grating

•Four regression methods are used to predict the load response of LPG sensor.•The proposed method is based on the fiber birefringence induced in LPG sensor.•Fourth-Degree Polynomial Fit can predict load intensity with the lowest error.•ANN was the best model to predict the azimuthal angle of applied...

Full description

Saved in:
Bibliographic Details
Published in:Measurement : journal of the International Measurement Confederation 2019-11, Vol.146, p.728-735
Main Authors: Barino, Felipe, Delgado, F.S., Jucá, Marco Aurélio, Coelho, Thiago V.N., Bessa dos Santos, A.
Format: Article
Language:English
Subjects:
Artificial neural networks
Birefringence
Correlation coefficients
Interrogation
Liquefied petroleum gas
Long-period grating
Neural networks
Optical fiber sensor
Optical fibers
Performance prediction
Polynomials
Regression analysis
Regression methods
Root-mean-square errors
Sensors
Strain analysis
Strain measurement
Strain rate
Studies
Transverse loads
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•Four regression methods are used to predict the load response of LPG sensor.•The proposed method is based on the fiber birefringence induced in LPG sensor.•Fourth-Degree Polynomial Fit can predict load intensity with the lowest error.•ANN was the best model to predict the azimuthal angle of applied load. In this work, we report the comparison of regression methods in a long-period grating (LPG) for transverse strain measurement. We analyze the transverse strain sensing characteristics, such as load intensity and azimuthal angle, based on the birefringence effect induced in LPG sensor. Therefore, we employ the different orthogonal responses of the grating to develop regression methods, which allow the estimation of the strain behavior of the LPG sensor. The predictive performances of these interrogation models are compared in terms of square correlation coefficient (R2) and root mean square error (RMSE). Finally, the results indicate that the best method to predict load intensity is the Fourth-Degree Polynomial Fit, whereas the artificial neural network (ANN) model could be successfully employed to predict the azimuthal angle.
ISSN:0263-2241
1873-412X
DOI:10.1016/j.measurement.2019.07.017