An uncertainty-based design optimization strategy with random and interval variables for multidisciplinary engineering systems

Abstract

Generally, the traditional uncertainty-based multidisciplinary design optimization (UBMDO) methods are based on the probability distribution information of random design variables and parameters. However, the probability distribution information needs to be obtained based on a large number of sample points. In many engineering structure design problems, for some uncertainties, it is difficult to obtain enough experimental samples to construct their accurate probability distribution, only the variation interval of them can be known. In this situation, mixed uncertainties exist in the engineering systems. Furthermore, with the increase in the complexity of engineering systems, these mixed uncertainties will even produce cumulative effects along with the internal coupling of the systems themselves. To tackle these challenges, the random and interval variables (RIV) are considered and a strategy of UBMDO with RIV (UBMDO-RIV) is proposed in this study. In the given method, the evaluation of uncertainty constraints is performed in the worst case scenario due to interval uncertainties. Meanwhile, the classic decoupling strategy for UBMDO, the framework of sequential design and uncertainty evaluation (SDUE), is introduced into UBMDO-RIV to reduce the computational burden. The engineering case study is utilized to illustrate the application of the proposed strategy.