Enhancing piezoelectric control of cracks in thin solid aluminum plates using finite element data and a neural networks approach

Smart material applications, such as piezoelectric actuators, have emerged as effective methods for repairing cracks in aircraft structures due to their electromechanical properties. Numerous studies have been conducted using finite element (FE) simulations; however, there is a lack of parametric studies utilizing advanced computing technologies. Therefore, this study focuses on the impact of bonding a piezoelectric actuator on mode I crack propagation in a thin-walled aluminum plate under plane stress conditions. Stress intensity factors (SIFs) are determined by using FE methods to examine the linear elastic fracture mechanics behavior at the crack tip. A comprehensive analysis is performed to evaluate the effects of actuator properties, adhesive characteristics, and plate dimensions. In addition, artificial neural networks, specifically feedforward neural networks, are employed for optimal SIF prediction, showcasing the ability to integrate soft computing with mechanical engineering challenges. To conduct the feedforward neural networks, the data have been prepared through FE simulations. These simulations have been executed considering the different parameters and their ranges. Later, these data have been arranged to train the algorithms. As a result, this combined approach highlights the feasibility of active crack repair using piezoelectric actuators for improved structural integrity. The results demonstrate that the integration of piezoelectric actuators effectively reduces SIF and provides real-time, data-driven optimization for crack management in thin-walled structures.