Abstract:The pressure and temperature of seawater are two important parameters. At present, people mainly rely on various types of temperature and depth measurement instruments to monitor the temperature and pressure of seawater. Optical temperature-depth (TD) sensors have advantages such as anti-electromagnetic interference, corrosion resistance and multiplexing capability. By using polydimethylsiloxane with a high thermo-optical coefficient and a high elastic-optical coefficient to encapsulate optical microfiber coupler combined sagnac loop (OMCSL) structure with large abrupt field characteristic, a simultaneous temperature and pressure fiber optic sensor with high stability and high sensitivity could be achieved. However, when using the conventional sensitivity matrix method (SMM) to demodulate the sensor, the demodulation results were unstable and encountered large error. One of the main reasons for the errors in the demodulation of the sensor using SMM is that the sensitivity matrix is an ill-conditioned matrix under certain conditions, and SMM in this state would greatly amplify the errors in the demodulation results. The other reason is that the feature wavelengths of the sensor would show nonlinear relationship with temperature when sensing in some environments. To reduce the demodulation error, in this paper, we researched and used various machine learning methods (MLMs) to demodulate the sensor. The experimental results showed that the demodulation error of the sensor could be greatly reduced by using the MLM compared to the traditional SMM. This paper provides a method to reduce the demodulation error of sensors with sensitivity nonlinear variation characteristics and reduce the demodulation error of sensors demodulated using SMM.

A rapid demodulation method for the interference spectrum of optical fiber sensors based on machine learning

Pattern Recognition Based on Enhanced Multifeature Parameters for Vibration Events in Φ‐otdr Distributed Optical Fiber Sensing System

Machine Learning Assisted Fast Demodulation of Large Dynamic Range Dual-Parameter Optical Fiber Sensors

A Novel Demodulation Scheme of Fiber-Optic Interferometric Sensor Based on FM SSB Signal

An Optical Orthogonal Heterodyne Demodulation Technology for Interferometric Fiber-optic Sensor

Large Dynamic Range Interrogation Technique for Fiber-Optic Interferometric Sensor Based on AWG and Deep Learning Algorithm

Application of machine learning in optical fiber sensors

High-speed spectrum demodulation of fiber-optic Fabry–Perot sensor based on scanning laser

Dual Parametric Simultaneous Demodulation of Fiber Optic Seawater Temperature and Pressure Sensors Based on Machine Learning Methods

Novel Demodulation Method for Fiber-Optic Interferometers Based on $\pi/2$ Phase Modulation

Fiber Interferometer Spectrum Recognition Strategy for Sensitivity Enhancement and Wide-Range Wavelength Measurements

Singlemode-Multimode-Singlemode Fiber-Optic Interferometer Signal Demodulation Using MUSIC Algorithm and Machine Learning

A New Demodulation Method to Improve the Sensitivity and Dynamic Range of Fiber Optic Interferometric System

Dynamic Demodulation of Low-Finesse Fabry-Perot Sensors Based on Instantaneous Frequency Analysis

Intelligent Sensing Analysis Using Mel-Time-Frequency-Imaging and Deep Learning for Distributed Fiber-Optic Vibration Detection

Federated learning-based wavelength demodulation system for multi-point distributed multi-peak FBG sensors

Machine learning enhanced smart tactile specklegram fiber sensor using spatial mode downsampling

A low cost all-digital demodulation method for fiber-optic interferometric sensors

A deep learning model enabled multi-event recognition for distributed optical fiber sensing

Multimode optical fiber sensors: from conventional to machine learning-assisted

A dynamic spectral demodulation scheme of the fiber optical Fabry-Perot sensor for low-frequency vibration detection