Predicting stress intensity factor in pressurized cracked tubes using neural networks

Artificial intelligence is fundamentally reshaping how we approach complex engineering challenges, offering solutions that are both swift and remarkably accurate. In this paper, we detail the development of a predictive model, built upon neural networks, to estimate the stress intensity factor (KI) in cracked tubes. By drawing upon a well-established analytical model, our aim was to devise a simple yet highly effective alternative to traditional numerical analyses. Thanks to strategic data augmentation, the model was meticulously refined and demonstrated exceptional accuracy. This study confirms AI’s potential to provide reliable and rapid modeling in the critical field of fracture mechanics. Moreover, the explicit nature of the mathematical equation resulting from our neural network allows for direct and easy implementation in computer or mobile applications, thus facilitating its practical use in the field.