Abstract
Optimization of many real-world optimization problems relies on numerical simulations for function evaluations. In some cases, both high- and low-fidelity simulations are available, where the high fidelity evaluation is accurate but time-consuming, whereas the low-fidelity evaluation is less accurate but computationally cheap. To find an acceptable optimum within a limited budget, it is economical for evolutionary algorithms to use both high- and low-fidelity evaluations in a single optimization search. This paper proposes a novel surrogate-assisted evolutionary algorithm using the transfer stacking technique for bi-fidelity optimization. To this end, a radial basis function network is firstly built to approximate the high-fidelity fitness function as additional low-fidelity evaluation, then a surrogate model transferring the original and additional low-fidelity evaluations to the expensive high-fidelity evaluation is adapted to guide the search. The simulation results on a series of bi-fidelity optimization benchmark problems with resolution, stochastic, and instability errors and a beneficiation processes optimization problem show that the proposed algorithm is both effective and efficient for solving bi-fidelity optimization problems, when their low-fidelity evaluations have resolution and stochastic errors. (C) 2020 Elsevier B.V. All rights reserved.