Minimizing Off-Axis Bending Effects on Flexible Surface Acoustic Wave Sensing Powered by Integrated Machine Learning Algorithms

Flexible surface acoustic wave (SAW) sensors have gained significant attention due to their favorable attributes such as conformability to curved surfaces, wireless/passive functions, and digital outputs. However, bending, especially complex off-axis bending deformation, often causes severe interference to the targeted detection signals with flexible SAW sensors, limiting their accurate monitoring on the curved/deformed surfaces. To address such a critical issue, we selected AlScN/ultrathin flexible glass-based SAW devices as an example, chose temperature as the targeted sensing parameter, and developed a model based on machine learning algorithms to minimize complex off-axis bending effects in temperature monitoring. Response characteristics of the flexible SAW devices to temperature variations and off-axis deformations were experimentally and theoretically investigated. Correlations between device’s responsive features and target parameter (temperature) were established using eight machine -learning algorithms. The optimized model was established with a normalized root mean square error of less than 1% and the determination coefficient R² was larger than 0.997 for temperature predictions subject to complex off-axis strain perturbations. Finally, the flexible SAW sensor showed a highly consistent temperature sensing capability under arbitrary off-axis bending conditions on a curved surface of a jet engine model.