Structural Damage Identification Using Multi-Objective Optimization Based Inverse Analysis

摘要

An optimization-based damage identification framework is proposed in this paper to analyze piezoelectric admittance data for structural health monitoring. We first establish finite element simulation of admittance information of piezoelectric transducer integrated to an underlying structure. Structural damage generally causes the change of admittance information which is then used to infer damage location and severity. In order to tackle the issue of under-determined problem in such an inverse analysis, we cast the damage identification process into an optimization problem to minimize the difference of between measured admittance change with respect to model prediction with assumed damage scenarios. Meanwhile, we take advantage the fact that the damage index vector is sparse in nature which corresponds to minimized l_0 norm. Subsequently, we formulate a multi-objective optimization problem to solve for the damage index vector. Dividing Rectangles algorithm is employed to solve the inverse identification problem, and three cases with different numbers of damage are considered and discussed. Results confirm that the proposed approach can deal with various damage scenarios.