An enhanced variable-fidelity optimization approach for constrained optimization problems and its parallelization

Abstract In this paper, a variable-fidelity constrained lower confidence bound (VF-CLCB) criterion is presented for computationally expensive optimization problems (COPs) with two levels of fidelity. VF-CLCB, the hierarchical Kriging model adopted to objective and inequality constraints. Two infill sampling functions are developed based on constraints, respectively, an adaptive selection strategy set select elite sample points. Moreover, VF-CLCB criterion, parallel method noted as PVF-CLCB subsequently accelerate process. PVF-CLCB, VF influence function defined approximately evaluate estimation error models, which multiple promising points can be determined at each iteration. addition, allocation proposed distribute computation resources between objective- constraint-oriented properly. Lastly, approaches compared alternative methods 12 benchmark numerical cases, their significant superiority in solving COPs verified. Furthermore, employed optimize global stability stiffened cylindrical shell, optimum structure yielded.