Reliability-based structural optimization using neural network

In this study two methods are examined. In the first one, the probabilistic perturbation method and the Edgeworth series are employed to give the theoretical formula for reliability-based optimization with non-normal random parameters. Therefore, the probabilistic constraints are transformed into deterministic constraints, and then the reliability-based optimal design parameters could be obtained accurately and quickly. In the second one, neural network (NN) and Monte Carlo simulation (MCS) are applied in reliability-based structural optimization with more failure modes. The neural network is used to simulate the relation expression between the design parameters and the structural reliability based on Monte Carlo simulation. Therefore, the structural probabilistic constraints are transformed into a single equivalent deterministic constraint approximately by NN, and then the process of reliability-based optimization can be implemented conveniently.