Data-Driven Modeling and Characterization of Carbon Nanotube Nanocomposite Strain Sensors for Human Health Monitoring Applications

Abstract

Developing electromechanical sensors with flexible and skin-mountable structures are highly demanded in human-related applications such as health care and sports fields. Conductive polymer nanocomposite with piezoresistive behavior is one solution to overcome stretchability and sensitivity limitations in conventional electromechanical sensors. In this study, we have developed a stretchable nanocomposite strain sensor based on multi-walled carbon nanotubes (MWCNTs) and Polydimethylsiloxane (PDMS) elastomer with improved behavior by investigating the effects of nanofillers concentration, curing temperature of the polymer, and the effect of mechanical pre-stretching on the response of the sensor. Optimizing these factors result in a significant improvement in the performance of the sensor by increasing the linearity (more than 20%), sensitivity (up to threefold), and hysteresis error decreased to less than 3%. However, the sensor still demonstrates relaxation errors during step strain and initial stretch/strain cycles test due to the viscoelastic nature of elastomers, which is common in every polymer-based nanocomposite sensor and, therefore, causes inaccuracy in practical applications. We developed a data-driven dynamic model using the System Identification technique in MATLAB to cover the time-dependent behavior of the sensor and predict the input strain more accurately. The model could reduce the strain prediction error up to 10% in comparison to the gauge factor equation. Finally, to demonstrate the potential applications of our sensor in human health applications, it was attached to the body for monitoring respiration and carotid artery signals.