Flexible-constrained time-variant hybrid reliability-based design optimization

Only the worst-case scenario is considered in most studies when conducting reliability-based design optimization under hybrid uncertainties including epistemic uncertainty and aleatory uncertainty, which will result in waste of resources because of the excessive pursuit of higher reliability. In order to quantitatively balance resources and reliability restricted by the lower and upper bounds under hybrid uncertainties during the design stage, a novel flexible-constrained time-variant hybrid reliability-based design optimization model is proposed in this paper. The infeasible region pruning-based Kriging method is proposed to build surrogate models for hard constraints while a combination of Kriging and high-dimensional model representation is presented to build surrogate models for flexible constraints to improve the efficiency. In order to build the relationship between resources and reliability, the determination method of design preference parameter is provided. A metaheuristic framework is finally given to conduct the flexible-constrained time-variant hybrid reliability-based design optimization. Two examples are employed to illustrate and validate the effectiveness of the proposed method.