Three different types (with glass, basalt and hybrid fibres) of composite rebars manufactured using the pultrusion process were loaded in four-point bending tests. All tests were carried out with acoustic emission sensors to better understand the mechanisms of damage. The data obtained were investigated using standard parameter analysis and also using unsupervised machine learning techniques called K-means. It was found that the best number of clusters is four or five. The numerical model using the finite-element method was calibrated on the basis of the experimental data. Further research will focus on numerical modelling of flexural behaviour of concrete beams reinforced with the presented composite rebars. The presented paper focuses on the characterization of the mechanical properties of composite rebars using a micromechanical approach, as well as analysis of progression damage processes appearing under flexural loading, using different perspectives provided by techniques such as acoustic emission analysis with machine learning-based clustering and numerical simulations. The presented research confirms that the proposed experimental-numerical approach can be applied in order to describe the flexural behaviour of Fibre Reinforcement Polymer (FRP) rods, which is relevant for investigating more complex cases of FRP concrete structures. This article is part of the theme issue 'Physics-informed machine learning and its structural integrity applications (Part 1)'.
Keywords: FEA; acoustic emission; clustering; composite rebars; k-means.