针对采用高斯过程进行建模时，不同核函数形式有着不同学习效果的问题。提出了一种自定义的平方指数形式的核函数，并基于多项式函数拟合对这种新形的核函数进行了数值仿真。仿真结果表明，该核函数不但可以提高模型的精确度和有效性，而且可以提高模型的学习能力和泛化能力。最后，将基于该核函数的高斯过程建模方法用于矩形双频微带天线优化设计和WLAN双频单极子天线优化设计，进一步证明了这种方法是可行的和有效的。 When modeling with Gaussian process, different kernel forms have different learning effects. A new self-defined kernel function in the form of square exponent is proposed. Based on the polynomial function fitting, the new kernel function is numerically simulated. The simulation results show that the kernel function can not only improve the accuracy and validity of the model, but also improve the learning ability and generalization ability of the model. Finally, the Gaussian process modeling method based on this kernel function is used to optimize the design of rectangular double-frequency microstrip antenna and WLAN dual-frequency monopole antenna, which further proves that this method is feasible and effective.