Interval multiobjective optimization of structures based on radial basis function, interval analysis, and NSGA-II

Abstract

To improve the multiple performance indices of practical engineering structures under uncertainties, an interval constrained multiobjective optimization model was constructed with structural performance indices included in objectives and constraints being functions of the interval uncertain parameters. An algorithm integrating radial basis function (RBF), interval analysis, and non-dominated sorting genetic algorithm (NSGA-II) was put forward to locate the Pareto-optimal solutions to the interval multiobjective optimization model. A series of RBFs were constructed based on the Latin hypercube experimental design (LHED) and finite element analysis (FEA), which were then integrated with interval analysis to compute the interval bounds of the objective and constraint functions under the fluctuation of uncertain parameters. Then the fitness of every individual during the NSGA-II-based optimization could be obtained. The case study on the optimization of the mechanical performance of a press slider with uncertain material properties demonstrated the feasibility and validity of the proposed methodology.