“…Improved Kriging function (i.e. effective expected feasibility function (EFF) (Bichon et al, 2012), learning function U (Zheng et al, 2017) and learning function H (Lv et al, 2015)), the best-expected sample points are precisely selected and sequentially added into the sample set in iterative process, the active learning function-based surrogate model can reduce the required sample number and improve the computational efficiency, whose basic thought is drawn in Figure 7 ( Wei et al, 2020). Combining the human-like inference ability of fuzzy logic system with the flexible structure of ANN model, a high-precision fuzzy neural network model method is proposed (Song et al, 2018) to address the complex multiple fluid-structure interaction calculation issues, as shown in Figure 8; By combining wavelet basis function with ANN structure, a wavelet neural network surrogate model method is proposed (Song et al, 2019b) to accomplish the fatigue reliability estimation, as shown in Figure 9; For the reliability analysis problem of complex time-varying environmental load, large time-varying characteristics and high instability of output response in time domain, an extreme response surface model method is proposed (Song et al, 2017) to transform the time-varying Reliability analysis of aeroengine rotor system process to output variables, as shown in Figure 10.…”