第 56 卷第 24 期期 220 they can make a trade-off between high prediction accuracy and low computational cost by augmenting the small number of expensive high-fidelity (HF) samples with a large number of cheap low-fidelity (LF) data. This work summarizes the state-of-the-art of MF surrogate modeling approaches and their applications in engineering design optimization. Firstly, the concept of three types of commonly used MF surrogate models is provided and the developments of extensions of them are reported. Secondly, the design of experiments for the MF surrogate models are summarized, including the one-shot design and sequential design approaches. Thirdly, two model management strategies, which directly determine the accuracy and efficiency of MF surrogate model-assisted design optimization approaches, are presented. Besides, the hot topics, MF surrogate model-assisted intelligent optimization algorithms and reliability/robust optimization are discussed. Fourthly, the applications of MF surrogate models in the practical engineering design domain are summarized. Finally, some suggestions about the usage of the MF surrogate models and their applications are provided, followed by the discussion of the deserved future work.