Estimation of Safety Index of Aging Structures through Neural Networks and Particle Swarm Optimization

Aging aerospace structures undergo severe damage due to corrosion, fatigue, creep, and other factors. The objective of this study is to estimate a unique parameter, called a safety index that gives an indication of the safety or reliability of these structures. To obtain the safety index, evolutionary computations using neural network based particle swarm optimization technique were developed to evolve the required data for the network. A simulation program, based on the neural network and optimization techniques, was implemented in MS visual basic/VC ⁺⁺ environment to predict the safety index of a damaged structure. Analytical solutions were used to predict the effects of crack, fatigue, and creep damage mechanisms. The results obtained from the simulation program were compared to the analytical solutions as there is no experimental data available for multiple damage mechanisms of an aging structure. Based on the results obtained, it appears that the proposed approach of obtaining the safety of a structure is reasonable, and the developed simulation program might be useful as an initial estimate to assess the structural integrity. However, more experimental work will be needed to establish the utility of this approach for practical applications.