NOVEL MICROWAVE SENSOR FOR ENHANCED BIOCHEMICAL DETECTION AND PREDICTION THROUGH MACHINE LEARNING FOR INDUSTRIAL APPLICATIONS

This paper presents a novel sensor design that incorporates a microstrip patch antenna accompanied by a ground plane integrating a complementary split-ring resonator (CSRR). Integration of a circular CSRR into the microchip antenna has the potential to significantly improve radiation characteristics. The designed sensor operates at a frequency of 2.45 GHz, achieving an attenuation level of -27 dB. This design proposes the sensor's potential to function as a highly sensitive sensor by utilizing changes in the dielectric constant of biological samples. The changing dielectric constant of the analyte induces a frequency shift, allowing for the identification of different materials. Additionally, various regression algorithms based on machine learning have been employed to accurately assess the analyte's dielectric constant by studying the sensor's frequency response. Performance analysis indicates that exponential regression outperforms other approaches, showcasing a minimal root mean squared error of 0.0013. Machine learning techniques bring about substantial enhancements in sensor performance, thereby creating pathways for sophisticated applications in biochemical sensing.