High-precision temperature-strain composite sensing based on a single resonator-type surface acoustic wave sensor using machine learning algorithms

Surface acoustic wave (SAW) sensors are extensively employed for temperature/strain measurement. However, precise temperature/strain sensing remains challenging under complex conditions involving simultaneous mechanical strains and temperature fluctuations, due to the inherent cross-sensitivity between temperature and strain. To address this challenge, we developed a single resonator-type SAW sensor operable up to 600 °C and integrated machine learning techniques for temperature-strain multiphysics decoupling. Six machine learning models are implemented and evaluated using normalized root mean square error (NRMSE), root mean square error (RMSE), mean absolute error (MAE), coefficient of determination (R²), and execution time (ET) metrics. Results demonstrate that the extreme gradient boosting (XGBoost) model achieves optimal predictive performance from 22 to 160 °C and applied strains of 0 με to 800 με. For temperature/strain predictions respectively, R² values reach 99.97% and 99.23%, simultaneously enabling high-precision temperature and strain sensing. Our study presents an effective solution for simultaneous temperature and strain sensing using a single resonator-type SAW sensor operating in a single acoustic wave mode.