Sizing of cavity defect in metallic foam from DC potential drop signals with stochastic inversion methods

Abstract

This paper focuses on the sizing of cavity defect in metallic foam (MF) from DC potential drop (DCPD) signals with stochastic optimization methods. Reconstruction schemes and numerical examples for reconstruction of the major defect parameters from both simulated and measured DCPD signals are given for the genetic algorithm (GA), the simulated annealing (SA) method, and the tabu search (TS) method respectively. The maximum reconstruction errors are 5.25% for the SA method with numerical signals, and 10.0% for the TS method with experimental signals, while the errors between the signals of MFs with defect of true parameters (or the experimental signals) and reconstructed parameters are small. This means that taking the potential drop signals of the nodes at the midline of the upper surface or the midlines of both the upper surface and the front surface is feasible to reconstruct the major parameters of single cavity defect with these stochastic methods. In addition, the numerical results suggest that the GA is the most efficient one among the selected stochastic inversion methods in terms of both accuracy and robustness, and the SA method is more time-saving.