Machine Learning-Enhanced Microfluidic SIW Sensor for Water-in-Fuel Detection Applications

This paper presents a half-mode substrate-integrated waveguide (HMSIW) sensor with a microfluidic integrated channel for detecting microliter water-in-diesel at 7.1 GHz. A microfluidic channel is embedded into the HMSIW structure, allowing interaction with the fields of TE
103
mode. The sensor's performance is assessed by following the resonance frequency fluctuations as water concentration varies in a
$65\ \mu\mathrm{L}$
sample using an examination of dielectric permittivity behavior. The resonance frequency is highly sensitive, shifting by 6.4 MHz per 0.05% water content. A back-propagation artificial neural network (BP-ANN) model is also deployed to allow cognitive capabilities, attaining a stunning detection range down to 0.01%. This sensor offers a potential option for real-time monitoring and quality control in fuel systems because to its multipurpose capabilities.