Identification of the interfacial cohesive law parameters of FRP strips externally bonded to concrete using machine learning techniques

A machine learning-based artificial neural network (ANN) approach is developed to automatically identify the interfacial cohesive parameters between fiber-reinforced polymers (FRPs) and concrete. A refined finite element (FE) model employing a cohesive zone model is established to simulate the interfacial Mode-II fracture. According to the database of load?displacement responses generated from the FE model, the trained ANN model can accurately and concurrently identify the cohesive law parameters. Moreover, based on a finite set of training data, the proposed approach shows high accuracy for the cases whose interfacial properties fall within the gap in or outside of the training dataset.

Automated summary

The study presents a machine learning-based artificial neural network approach to automatically identify the interfacial cohesive parameters between fiber-reinforced polymers and concrete. By utilizing a refined finite element model with a cohesive zone model and a trained ANN model, the cohesive law parameters can be accurately identified, showing high accuracy even for cases falling outside the training dataset gap.