Applications of Optimization Algorithms for Structural Damage Detection

Abstract

Applications of optimization algorithms along with finite element (FE) software have become a promising field in civil engineering, predominantly for structural monitoring, since they attenuate the need for regular inspection campaigns, minimizing time, transport costs, and potential human errors. Hence, this article intends to evaluate the efficiency of optimization algorithms, specifically, the Genetic Algorithm (GA) and the Particle Swarm Optimization (PSO), for locating and quantifying the damage in steel and concrete frames. Therefore, numerical models of deteriorated concrete beams (BEAM188) were developed in ANSYS Mechanical APDL. Furthermore, modal results, which comprise natural frequency and vibration modes, were extracted from FE to calibrate the structures with GA and PSO, implemented in Python, aiming to obtain the damaged coordinate and the degree of deterioration for the girder. GA converged well, showing accuracy, when evaluating the coefficients of variation obtained: 0.03 and 0.15, for position and intensity of damage, respectively. PSO displayed good results, but in one case, this technique failed. In addition, it was noted that the efficiency of the algorithm decreased when the damage was small. Thus, the algorithms proved capable of identifying structural damage, despite PSO failing to escape local minima.