Fabrication and Characterization of a Microwave Sensor for Potassium Hydroxide Analysis and prediction Using Gaussian Process Regression

This study presents the design and fabrication of a microwave sensor for the precise detection of potassium hydroxide (KOH) solution concentrations, ranging from 0.52mol/kg to 3.22mol/kg. The sensor incorporates a novel mirrored E-shaped metamaterial cell, leveraging its distinctive magnetic properties to enhance sensor performance. To validate the design, we conducted comprehensive performance evaluations using both experimental measurements and simulations, employing the Finite Integration Technique (FIT), Finite Element Method (FEM), and Method of Moments. The electrical properties of the KOH solutions were accurately characterized using the Cole-Cole model. Experimental results revealed a significant shift in the sensor's resonance frequency (Δ f ), reaching a maximum of 0.52 GHz, while maintaining high amplitude sensitivity (\(\:\left|{\Delta\:}{S}_{21}\right|=0.57\:dB,\:\left|{\Delta\:}{S}_{11}\right|=1.63\:dB\)) within the 1–5 GHz frequency band. Furthermore, we observed notable variations in the sensor's quality factor (Q), ranging from 11.25 to 0.42. These variations directly correlated with changes in the KOH solution concentration, demonstrating the sensor's high sensitivity. To further analyze the sensor's behavior, we utilized a Gaussian Process Regression model to predict the reflection coefficient (S ₁₁ ) and transmission coefficient (S ₂₁ ) of the KOH solution. The predicted S-parameter values exhibited a close agreement with the experimental measurements, achieving a low Mean Absolute Percentage Error (MAPE) of approximately 4.31% for S ₁₁ and 1.90% for S ₂₁ at a target concentration of 2.64mol/kg.