Integrating Swarm Intelligence with Neural Networks: A Combination Approach for Predicting Beam Cracks

Detecting and locating damage is a crucial aspect of structural health monitoring. While Artificial Neural Networks (ANNs) have shown success in identifying damage in civil and mechanical structures, they come with certain limitations. However, enhancing the effectiveness of ANNs is achievable through adjustments in their architecture and training strategies. This study introduces a metaheuristic algorithm, specifically the Butterfly Optimization Algorithm (BOA), to optimize an ANN for predicting multiple damages in aluminum bars. Input parameters include natural frequencies, and output parameters consist of crack depths. The paper employs an enhanced Finite Element Model (FEM) to gather data through simulation, considering various crack depths. To gauge the dependability of this method, we gather experimental data from the examination of beams with varying crack depths. The results obtained are juxtaposed with comparable approaches employing metaheuristic algorithms like the Artificial Bee Colony Algorithm (ABC) and Genetic Algorithm (GA). The newly proposed approach demonstrates robust performance in predicting damage, showcasing its efficacy in comparison to alternative methods.