Real-time in-depth damage identification and health index system for carbon fiber-reinforced composites using electromechanical behavior and data processing tools

Structural health monitoring using electromechanical behavior can help detect various damage types and failure modes in composites. However, only the presence of damage and structural failure can be monitored. For a thorough identification of damage in composites, this paper proposes an electromechanical data analysis and processing methodology using principal component analysis and k -means clustering. The health state of unidi-rectional carbon fiber-reinforced plastic (CFRP) composites was monitored using self-sensing data. Various types of damage and failure modes in carbon fibers with different directionality were investigated based on in-depth damage analysis using a machine-learning-based data processing technique. A novel health index system for damage propagation investigation was proposed based on an electromechanical behavior analysis. The results produced by the damage index system were compared with those obtained by ABAQUS simulation and me-chanical behavior analysis to determine the rationality of the system. An advanced condition-based monitoring methodology can help investigate the current health state of composites and the propagation of different types of damage. The proposed system has potential applications, and our results provide guidelines for self-sensing research.

Automated summary

This paper proposes an electromechanical data analysis and processing methodology using principal component analysis and k-means clustering for a thorough identification of damage in composites. Various types of damage and failure modes in carbon fibers with different directionality were investigated using a machine-learning-based data processing technique. A novel health index system for damage propagation investigation was proposed based on an electromechanical behavior analysis. The proposed system has potential applications and provides guidelines for self-sensing research.