Intensity and Wavelength-Division Multiplexing Fiber Sensor Interrogation Using a Combination of Autoencoder Pre-Trained Convolution Neural Network and Differential Evolution Algorithm

Po-Han Chiu,Yibeltal Chanie Manie,Ja-Hon Lin,Yu-Shen Lin,Jyun-Wei Li,Peng-Chun Peng

doi:10.1109/jphot.2021.3050298

Po-Han Chiu, Yibeltal Chanie Manie + Show 4 more

Open Access

https://doi.org/10.1109/jphot.2021.3050298

Copy DOI

Abstract

This paper proposes a new fiber Bragg grating central wavelength interrogation system by combining evolutionary algorithm and machine learning techniques integrated with an unsupervised autoencoder (AE) pre-training strategy. The proposed unsupervised AE pre-training convolution neural network (CNN) allows training of the convolutional layers independently from a regression task in order to learn a new data representation and give better generalization. It is also used to improve the system accuracy by four times without extra-labeled data. Moreover, AE is combined with a differential evolutionary (DE) algorithm to automate the human labeling task. The proposed autoencoder pre-training convolution neural network and differential evolutionary (AECNNDE) interrogation system achieve good accuracy and can speed-up the computational time by a maximum of 30 times than DE algorithm.

Highlights

The fiber sensor has advantages in multiple measurement fields to measure different parameters [1]–[6]
This paper proposes a new fiber Bragg grating central wavelength interrogation system by combining evolutionary algorithm and machine learning techniques integrated with an unsupervised autoencoder (AE) pre-training strategy
Where λN,m is the central wavelength that depends on the measurands, Ip is the reflected peak intensity, λB represents the full width at half maximum (FWHM) of the Fiber Bragg grating (FBG)

Summary

Introduction

The fiber sensor has advantages in multiple measurement fields to measure different parameters [1]–[6]. The reflection spectra of cascading FBG sensors cannot overlap in the WDM system, which limits the number of sensors multiplexed in the sensor system To overcome this problem and to further increasing the number of sensors in the system, the intensity and wavelength division multiplexing (IWDM) method was proposed [16]–[19]. In evolutionary algorithms, the computational cost increases exponentially when the number of FBGs in the sensor system increase To solve this issue, machine learning-based central wavelength interrogation algorithms are proposed to provide high-speed inferences and parallel computing. In the previously proposed convolutional neural network (CNN) based machine learning interrogation system [21], 24000 samples are used to train the model This kind of training data is often hard to prepare for a real sensing scenario and only fits into one case.

Operation Principle and Experiment Setup

Interrogation System

Experiment Result

Conclusion