Anti-noise proxy optimization method capable of reducing numerical noise influence in design

The invention provides an anti-noise proxy optimization method capable of reducing numerical noise influence in design. The anti-noise proxy optimization method comprises the following steps: determining an optimization design problem; constructing a sample set; establishing a support vector regression model of the design variables about the target function; optimizing by taking the sample set as a training sample to obtain an optimal value of a hyper-parameter epsilon; according to the optimal value of the hyper-parameter epsilon, taking the sample set W as a training sample, and constructing an epsilon-Kriging proxy model; and adding points by adopting a constraint EI function point adding method until a result meeting an optimization design target is obtained. The invention provides an anti-noise agent optimization method capable of reducing the influence of numerical noise in design, the influence of the numerical noise in optimization design can be filtered out, and the optimization result meeting the de