A Data Fusion Framework for Fracture Toughness Modeling Using Multiple Sources of Data

The adhesive bonding technology of composite material is widely used in the industry, and the double-cantilever beam (DCB) test is a standard test for measuring the bonding quality. However, adhesive bonding methods may compromise the bonding strength, leading to weak bonds or so-called kissing bonds. In this research, we present a data-driven method to model the relationship between the process parameters and the mode-I fracture toughness. Due to the limited size of the DCB training data, we propose a novel data fusion framework, also incorporating the historical single-lap joint (SLJ) dataset at hand. Though the SLJ test is a less effective method for measuring the fracture toughness, we show it can be used to improve the model performance. We then demonstrate the effectiveness of our data-driven framework in an airplane maintenance application, with two times better predictive performance obtained.