Machine learning-enhanced surface plasmon resonance based photonic crystal fiber sensor

Abstract

This study introduces what we believe to be a novel photonic crystal fiber sensor utilizing surface plasmon resonance (SPR), incorporating four gold nanowires to enhance sensing capabilities. The research employs machine learning, specifically artificial neural networks (ANN), to predict confinement loss and sensitivity, achieving high accuracy without needing the imaginary part of the effective refractive index. The machine learning technique is applied in three different scenarios, resulting in mean squared errors of 0.084, 0.002, and 0.003, highlighting the reliability of the ANN models in predicting sensor outputs. Additionally, the sensor demonstrates impressive wavelength sensitivities of 2000-18000 nm/RIU (nanometers per refractive index unit) for refractive indices of 1.31-1.4 within the 720-1280 nm wavelength range, and a notable maximum amplitude sensitivity of 889.89 RIU-1. This integration of SPR, photonic crystal fiber, and machine learning not only optimizes sensor performance but also offers an efficient methodology for prediction, showcasing the potential of machine learning in advancing optical sensor design.