Wonnacott Alex M, Bowden Anton E, Mitchell Ulrike H, Fullwood David T
Department of Mechanical Engineering, Brigham Young University, Provo, UT 84602, USA.
Department of Exercise Sciences, Brigham Young University, Provo, UT 84062, USA.
Sensors (Basel). 2024 Dec 22;24(24):8192. doi: 10.3390/s24248192.
Flexible high-deflection strain gauges have been demonstrated to be cost-effective and accessible sensors for capturing human biomechanical deformations. However, the interpretation of these sensors is notably more complex compared to conventional strain gauges, particularly during dynamic motion. In addition to the non-linear viscoelastic behavior of the strain gauge material itself, the dynamic response of the sensors is even more difficult to capture due to spikes in the resistance during strain path changes. Hence, models for extracting strain from resistance measurements of the gauges most often only work well under quasi-static conditions. The present work develops a novel model that captures the complete dynamic strain-resistance relationship of the sensors, including resistance spikes, during cyclical movements. The forward model, which converts strain to resistance, comprises the following four parts to accurately capture the different aspects of the sensor response: a quasi-static linear model, a spike magnitude model, a long-term creep decay model, and a short-term decay model. The resulting sensor-specific model accurately predicted the resistance output, with an R-squared value of 0.90. Additionally, an inverse model which predicts the strain vs. time data that would result in the observed resistance data was created. The inverse model was calibrated for a particular sensor from a small amount of cyclic data during a single test. The inverse model accurately predicted key strain characteristics with a percent error as low as 0.5%. Together, the models provide new functionality for interpreting high-deflection strain sensors during dynamic strain measurement applications, including wearables sensors used for biomechanical modeling and analysis.
柔性高挠度应变片已被证明是用于捕捉人体生物力学变形的经济高效且易于获取的传感器。然而，与传统应变片相比，这些传感器的解读要复杂得多，尤其是在动态运动过程中。除了应变片材料本身的非线性粘弹性行为外，由于应变路径变化期间电阻出现尖峰，传感器的动态响应更难捕捉。因此，从应变片电阻测量中提取应变的模型通常仅在准静态条件下才能很好地工作。本研究开发了一种新颖的模型，该模型能够捕捉传感器在周期性运动过程中完整的动态应变 - 电阻关系，包括电阻尖峰。将应变转换为电阻的正向模型包括以下四个部分，以准确捕捉传感器响应的不同方面：准静态线性模型、尖峰幅度模型、长期蠕变衰减模型和短期衰减模型。所得的特定于传感器的模型准确预测了电阻输出，决定系数R²值为0.90。此外，还创建了一个逆模型，该模型预测会导致观察到的电阻数据的应变与时间数据。逆模型是在单次测试期间根据少量循环数据针对特定传感器进行校准的。逆模型以低至0.5%的百分比误差准确预测了关键应变特征。这些模型共同为动态应变测量应用（包括用于生物力学建模和分析的可穿戴传感器）中解读高挠度应变传感器提供了新功能。
