Structural Damage Identification Using Data Fusion and Optimization of the Self-Adaptive Differential Evolution Algorithm

This paper addresses the critical challenges of inadequate localization and low quantification precision in structural damage identification by introducing a novel approach that integrates Dempster–Shafer (D-S) evidence theory with the Self-Adaptive Differential Evolution (SDE) algorithm. First, modal parameters are extracted from a simply supported beam using the finite element (FE) method, and the corresponding index values are computed based on the formulated damage identification index equations. Next, these indices are applied to analyze damage localization in both single-position and multi-position scenarios within the simply supported beam. The SDE algorithm is then employed to dynamically optimize the initial weights and thresholds of various algorithms, ensuring the assignment of optimal values. Finally, the resulting data are input into the model for training, yielding a prediction model with enhanced accuracy that can precisely estimate the damage severity of the simply supported beam. The findings demonstrate that the three proposed damage identification indices—DI1,i,j, DI2,i,j, and DSDIi,j—not only achieve high accuracy in damage localization but also significantly improve the precision of algorithms optimized by the SDE. These methods exhibit strong accuracy and robustness, providing a valuable reference for damage identification in small-to-medium-span simply supported beam bridges.