Damage detection for a large-scale truss bridge using Tranmissibility and ANNAOA

In this paper, we propose an efficient approach to enhance the capacity of Artificial Neural Network (ANN) to deal with Structural Health Monitoring (SHM) problems. Over the last decades, ANN has been extensively utilized for damage detection in structures. In order to identify damages, ANN frequently utilizes input information that is based on dynamic features such as mode shapes or natural frequencies. However, this type of data may not be able to detect minor damages if the structural defects are insignificant. To transcend these limitations, in this work, we propose utilizing transmissibility to create input data for the input layer of ANN. Moreover, to deal with local minimum problems of ANN, a combination between the Arithmetic Optimization Algorithm (AOA) and ANN is proposed. The global search capacity of AOA is employed to remedy the local minima of ANN. To evaluate the effectiveness of the proposed approach, a numerical model with different damage scenarios is considered. The suggested approach detects damage location precisely and with higher severity detection precision than the conventional ANN method.