基于遗传进化的支持向量机参数敏感性研究

提出了一种基于遗传进化的支持向量机预测方法,利用该预测方法对工程实例阳宗隧道下行线XK38+725断面的实测围岩变形进行预测,发现此方法具有很高的精度,同时,对基于遗传进化的支持向量机参数敏感性进行了研究,得出结论:敏感性最高的是不敏感系数,其次是惩罚因子,敏感性最低的是核函数系数。

Based on the genetic evolution of the support vector machine parameter sensitivity study

In this paper,raising that based on the genetic evolution of support vector machine prediction method,using the prediction method to Yangzong tunnel project XK38+725 descending line of the measured cross-section to predict rock deformation,finding that this method has high accuracy,at the same time,based on genetic evolution of the support vector machine parameter sensitivity is studied,concluded that: the highest sensitivity is not sensitive coefficient,followed by the penalty factor,the lowest sensitivity is coefficient of nuclear function.