k-Means Clustering for Prediction of Tensile Properties in Carbon Fiber-Reinforced Polymer Composites

The application of computer algorithms to identify patterns in data is referred to as machine learning. The algorithms are used to learn complex relationships and build models for various predictions. Herein, the k-means method is used, one of the unsupervised learning methods in machine learning, to predict Young's modulus and ultimate tensile strength (UTS) of carbon-fiber-reinforced polymers (CFRPs), and their experimental Young's modulus and UTS values are compared. The k-means method categorizes CFRP into four colors: carbon fiber, epoxy resin matrix, defects, and contamination. The prediction of Young's modulus and UTS of CFRP with different porosities and carbon fiber orientation demonstrates the effectiveness of the k-means method. Furthermore, the experimental values of Young's modulus and UTS of commercial CFRP plate are closer to the predicted values than the catalog values. These results suggest that the k-means method can predict Young's modulus and UTS of CFRP accurately, instantly, and automatically. The k-means method is promising as a new technique to accurately and instantly understand the mechanical and physical properties of CFRPs without any material test.