Investigate of Temperature and Strain Effect on Fiber Grating Performance: A Mathematical Approach for Data Generation

Wong, Cheng En (2026) Investigate of Temperature and Strain Effect on Fiber Grating Performance: A Mathematical Approach for Data Generation. Final Year Project (Bachelor), Tunku Abdul Rahman University of Management and Technology.

Text WONG CHENG EN_Full Text.pdf Restricted to Registered users only Download (6MB)

Abstract

This dissertation investigates the impact of temperature and strain on uniform Fiber Bragg Grating (FBG) sensors through a hybrid simulation and deep learning framework, addressing the limitations of costly experimental setups and restricted real-world data collection. FBG sensors, known for their electromagnetic immunity and durability in harsh environments, exhibit Bragg wavelength shifts under external stimuli, making them valuable in structural health monitoring, aerospace, and biomedical applications. To simulate these behaviours, transformed Coupled Mode Theory (CMT) was implemented in MATLAB R2023a, enabling automated generation of 26,888 synthetic spectra whereby 14,880 temperature-variant from –27 °C to 73 °C, 0.1 °C steps and 15 spans;12,008 strain-variant from 2000 to 5000 με, 1 με steps and 4 spans were generated, replicating acrylate FBG properties with sensitivities of 13.7 pm/°C and 1.21 pm/με. Datasets were exported in Portable Network Graphic (PNG) and (Comma-Separated Values) CSV formats, preserving spectral integrity and confirming linear responses. Pre-processed spectra were segmented in 70:15:15 split and classified using a fine-tuned ResNet50 convolutional neural network (CNN), achieving 99.58% test accuracy on four classes at a 0.0001 learning rate, with metrices achieving over 0.99 and class-wise accuracies of 98.39 to 100% across 4 classes, outperforming traditional methods while mitigating cross-sensitivity. The contributions include a scalable, cost-effective alternative to hardware-based experiments, validated benchmarks for FBG response, and an Artificial Intelligence (AI)-driven framework aligned with Sustainable Development Goals (SDG) 9: Industry, Innovation, and Infrastructure. Future work should extend validation with real-world spectra, explore multi-parameter sensing, and advance deployment through modern architectures such as Vision Transformers (ViT)