v支持向量回归用于退化轨迹建模

针对小样本情形下的退化轨迹建模问题,为解决用ε-支持向量回归(ε-support vector regression,ε-SVR)建模时不敏感参数ε不易选择的难题,提出一种基于v支持向量回归(v-support vector regression,v-SVR)的退化轨迹建模方法,并用遗传算法优化模型参数以提高建模精度。参数v-与支持向量和错误样本点的个数有关,根据这一性质确定v的取值范围,并实现对支持向量或错误样本点个数的控制。对疲劳裂纹增长数据的实例分析表明,所提方法不仅便于确定参数,而且相对于以往文献的方法有更高的建模精度。

Modeling degradation path using v-support vector regression

In modeling degradation path with small sampling, the parameter ε of theε-support vector regression (ε -SVR) is difficult to select. In order to solve this problem, a degradation path modeling method is proposed, on the basis of v-support vector regression (v-SVR). The genetic algorithm (GA) is adopted to optimize parameters of the proposed model. The parameter v has property that it correlates with the number of support vectors and inaccuracy samples, so the range of v’s value can be determined, and v can be also used to control the number of support vectors and inaccuracy samples. The proposed method is applied to fatigue crack growth data, the result indicates that the proposed approach makes the selection of parameters much easier, and gains a higher modeling accuracy than some other methods in the existing literature.