High-accuracy real-time measurement by viscoelastic flexible composite sensors via electromechanical modeling and UKF data fusion

Abstract

Flexible strain sensors based on soft materials have been widely adopted in wearable applications recently. However, the inherent viscoelasticity of these materials inevitably leads to stress and resistance relaxation, which limits their use to more qualitative rather than quantitative. To address real-time measurement error, this work studied a typical conductive composite strain sensor made of carbon nanoparticle-filled silicone. First, a viscoelastic electromechanical model for this sensor was established, accompanied by an efficient parameter identification scheme. Subsequently, a multi-sensor angle measurement platform was designed and fabricated based on this sensor, using IMU gyroscopes for additional detection of strain velocity. Finally, by integrating the unscented Kalman filter with the proposed electromechanical model, high-accuracy joint angles were provided through data fusion. Results show that using the model and proposed fusion algorithm, joint angles can be accurately measured for both dynamic and quasi-static scenarios, achieving a high precision with a coefficient of determination R2 > 0.99 and a root mean square error < 2.28°. To conclude, this approach effectively compensates for the sensor’s intrinsic errors induced by rheology with electromechanical modeling and data fusion framework, thereby significantly enhancing measurement accuracy. The methods and techniques presented can inspire high-precision applications of other flexible sensors.