Water-Based SRR Sensor With Machine Learning Algorithms for Simultaneous Temperature and Pressure Detection

Water resources are vital for both human beings and the Earth’s ecosystem, serving as a crucial asset for survival while also being a low-cost, environmentally friendly resource. However, in the field of electronic devices, water often signifies corrosion and failure, and very few of studies has investigated the feasibility of developing water-based devices. In this study, a wireless and powerless water-based split ring resonator (WSRR) sensor with machine learning (ML) algorithms was developed for simultaneous temperature and pressure detection. The sensor is fabricated by filling conductive water into a flexible substrate with a resonant ring structure. Within the operational frequency range of the sensor, the dielectric properties of sensor materials vary with temperature, which in turn affects the sensor’s output. Through a combination of simulation and experimental methods, it has been demonstrated that the parameter <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\vert {S}_{{11}}\vert $ </tex-math></inline-formula> in the output frequency curve can be used to measure temperature changes. Additionally, the fluidity of the conductive water, coupled with the flexibility of the substrate material, allows environmental pressure to be converted into deformation of the conductive material. Similarly, we have verified that the parameter <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${f}_{r}$ </tex-math></inline-formula> in the output frequency curve can be used to detect pressure changes. The device has a temperature and pressure sensitivity of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$3.4\times 10^{-{5}}$ </tex-math></inline-formula> dB/°C and 83 kHz/kPa, respectively. ML algorithms are used to eliminate the mutual interference of dual-parameter measurements for temperature and pressure. The trained model achieves an accuracy of 98.95% in predicting pressure, and the accuracy for temperature prediction is 88.67%.