[Springer] AI-Driven Multi-Objective Optimization of an Au/MoS2/PVP-Coated PCF-SPR Biosensor for Glycerol Concentration Sensing Copy

The growing demand for highly sensitive, rapid, and label-free biosensing technologies has accelerated the development of Photonic Crystal Fiber-based Surface Plasmon Resonance (PCF-SPR) sensors for biochemical applications. In this study, an artificial intelligence AI-assisted multi-objective design and optimization is proposed to employ Large-Mode-Area Polarization-Maintaining PCF (LMA-PM-10 PCF) as the sensing platform due to its excellent polarization stability and optical confinement characteristics. To improve sensing efficiency and plasmonic interaction, a hybrid multilayer configuration consisting of gold (Au), molybdenum disulfide (MoS2), and polyvinylpyrrolidone (PVP) is incorporated into the sensor structure. The optical performance of and dataset for the proposed biosensor is numerically investigated using COMSOL Multiphysics. The refractive index (RI) of the analyte was varied between 1.33 and 1.40 to evaluate the sensing behavior over a broad range of biological and chemical environments. Key performance parameters, including wavelength sensitivity (Sλ), confinement loss (CL), resonance wavelength (λres), Full Width at Half Maximum (FWHM), and Figure of Merit (FOM), were investigated to evaluate sensor behavior. Subsequently, Artificial Neural Networks (ANN), Particle Swarm Optimization-assisted ANN (PSO-ANN), and the Non-dominated Sorting Genetic Algorithm II (NSGA-II) were employed to model, predict, and optimize the sensor response. In the standard optimization method, a sensitivity of 4462 nm/RIU was achieved; however; with the assistance of Machine Learning (ML) algorithms, this was increased to 5850 nm/RIU, representing a significant improvement of 31.1%. The FOM was significantly enhanced from 213 RIU− 1 to 295 RIU− 1.The PSO-ANN achieved an improvement in terms of Mean Square Error (MSE), estimated at 77.88%, compared with ANN. This study will considerably contribute to the literature in terms of establishing the optimum thicknesses of plasmonic and other functional materials for the proposed sensor type to improve sensor characteristics such as sensitivity, CL, FOM, FWHM and resonance wavelength, along with the selection of an accurate optimization algorithm type.