论文部分内容阅读
以凉山州安宁河流域129个乡镇的泥石流危险性区划资料为依据,随机选取总样本数的2/3和1/2作为训练样本,建立不同数量训练样本下安宁河流域泥石流危险性区划的多分类SVM模型,进行以乡镇为单元的区域泥石流危险性评价研究。评价结果表明,SVM模型的预测精度随着训练样本数量的增加而提高;2个SVM模型对测试样本的预测准确率均高于相应的BP神经网络模型,对训练样本的回判准确率高于或接近于BP神经网络模型。因此,支持向量机方法是一种比神经网络方法具有更优精度和更强泛化性能的新机器学习方法,在泥石流危险性评价实践中具有十分广阔的应用前景和推广应用价值。
Based on the data of debris flow risk in 129 towns and villages of Anning River basin in Liangshan Prefecture, two-thirds and one-half of the total number of samples were randomly selected as training samples to establish the multi-hazard regionalization of debris flow in Anning River Basin under different training samples The classification of SVM model, the township as a unit of regional debris flow risk assessment. The evaluation results show that the prediction accuracy of SVM model increases with the number of training samples. The prediction accuracy of the two SVM models is higher than that of the corresponding BP neural network model, and the accuracy rate of the training samples is higher than Or close to the BP neural network model. Therefore, the SVM method is a new machine learning method with better precision and more generalization than the neural network method. It has a very broad application prospect and popularization and application value in the practice of debris flow hazard assessment.