Impact damage characterization approach for CFRP pipes via self-sensing

Abstract

The growing usage of carbon-fiber-reinforced polymers (CFRPs) has necessitated the development of structural health monitoring (SHM) for ensuring their integrity and safe operation. This paper presents a novel technique for monitoring the structural health of 3D CFRP pipe structures subjected to multiple impacts based on real-time electrical resistance monitoring. Damage characteristics such as its location and type were identified using the monitoring data and machine learning techniques. Additionally, finite element analysis was deployed to rationalize the electromechanical behaviors of pipes with varying stacking types and thus different fracture mechanisms. Analysis of the electrical resistance data coincided with the results of the finite element analysis in terms of damage modes and types. Therefore, a simple real-time electrical resistance measurement can detect the damage type without numerical analysis. A damaged section was identified by comparing the intensity of the data. Through machine learning techniques, the timepoints of damage severity demarcation were accurately predicted within error of 10 s and accuracies over 96 %. The machine-learning tools applied are unsupervised, thus reducing the effects of data dependency, labeling, and imbalance, as well as ensuring flexibility and general applications. The proposed technique was able to detect the damage location, severity, and type in real-time; hence, it has various potential applications in structural health monitoring including CFRP materials and is one of the few methods used to assess 3D structure characterization using self-sensing, which can result in mature technology and the optimal operation of CFRP systems.