Machine learning (ml) approaches to model interdependencies between dynamic loads and crack propagation

The application of machine learning in structural health and crack prediction is of paramount importance, as it offers the potential to enhance the accuracy, efficiency, and reliability of detecting and predicting damage in various materials and structures. This research presents an in-depth exploration of machine learning (ML) applications in the field of Structural Health Monitoring (SHM) across various materials, including composites, metals, and polymers. The study identifies the current challenges in implementing ML in SHM, such as data sparsity, interpretability of ML models, overfitting, and the absence of general guidelines for ML model selection. The research analyses the dynamic response data of different materials and establishes significant crack depth predictors for materials such as aluminum, concrete, and 3D-printed Acrylonitrile Butadiene Styrene (ABS). It further investigates and validates selected ML models to predict crack depth in different materials. The models' performance is evaluated using Mean Squared Error (MSE) on both training and test sets, demonstrating their ability to capture meaningful patterns within the data and make reasonably accurate predictions. A significant contribution of this study is the proposal of an automated model utilizing the H2O library for crack propagation prediction in ABS materials. This model demonstrates the potential of automation in SHM, offering substantial benefits for structural integrity assessment, maintenance strategies, and materials design in various industries. This research concludes with recommendations for future research, including the exploration of advanced ML algorithms, investigation of additional predictive features, and evaluation of the models in different real-world scenarios.