Identifying Damage in ۳D Skeletal Structures Using a Two-Stage Probabilistic Approach by the MTLBO Optimization Algorithm

This paper introduces an advanced structural health monitoring (SHM) framework that integrates Bayesian data fusion with a Modified Teaching–Learning-Based Optimization (MTLBO) algorithm to improve damage identification in space frames and truss structures. The proposed method utilizes multiple vibration-based damage indices, including Modal Strain Energy (MSE), Frequency Response Function Strain Energy (FRFSE), Flexibility Strain Energy (FSE), and Residual Force-Based (RFB) indicators, derived from natural frequencies and mode shapes. These indices collectively enhance damage localization accuracy while reducing the number of false or redundant damaged elements, thereby improving computational efficiency. Through the application of Bayesian data fusion, the approach establishes a robust probabilistic framework that effectively integrates different damage indicators to refine detection results. The incorporation of the MTLBO algorithm further enhances the convergence speed and accuracy of the optimization process, expediting the identification of damage across complex structural systems. Moreover, by employing a probabilistic objective function, the framework demonstrates notable resilience to measurement noise, ensuring reliable performance under uncertain conditions. A key innovation of this study lies in the simultaneous probabilistic treatment of both damage localization and damage severity estimation, enabling more precise, noise-resistant, and computationally efficient damage detection. This hybrid methodology represents a significant advancement in SHM, offering a robust and practical solution for real-world structural assessment and maintenance applications.

کلیدواژه ها:

Bayesian fusion ، Damage detection ، Modify teaching learning-based ، Multi-damage detection ، Structural health monitoring