支持向量机在边坡稳定性评价中得到了较为广泛的应用,但目前学者们对边坡的稳定性等级的分级数量及分级区间尚未有统一的认识,同时对训练样本数和验证样本数的划分亦停留在经验层面,较少探讨其最优的样本数。为此选取边坡的坡高、坡角、容重、粘聚力以及内摩擦角五个参数为评价指标,使用GeoStudio软件对不同评价指标的组合进行数值模拟,得到963组数据。使用MATLAB软件对该数据进行SVM模型训练分析后认为边坡的稳定性等级可分为8个等级,验证样本平均准确率为88.39%;当样本总量固定时,验证准确率随着训练样本数的增大而增大;当无其他划分经验参考时,尽量让每个分类等级的验证样本数量约为12个。

Support Vector Machine (SVM) has been widely applied in slope stability assessment. However, currently, scholars have not reached a consensus on the number and interval of slope stability levels. Moreover, the division of training and validation samples remains largely based on empirical knowledge, and the optimal sample size has received less exploration. To address this, we selected five parameters, including slope height, slope angle, bulk density, cohesion, and internal friction angle, as evaluation indicators for slope stability. We conducted numerical simulations using GeoStudio software to obtain 963 sets of data for different combinations of evaluation indicators. The data was then analyzed and trained using MATLAB software to develop an SVM model. The analysis suggested that slope stability levels can be divided into 8 categories, with an average accuracy rate of 88.39% for the validation samples. Furthermore, as the number of training samples increases, the accuracy of the validation also improves, assuming a fixed total sample quantity. Additionally, in the absence of other partitioning references, it is advisable to have approximately 12 validation samples for each classification level.