针对传统泥石流危险性评价方法存在的缺点,本文运用支持向量机理论,以云南省30条泥石流沟和黄河积石峡的10条泥石流沟的数据为样本进行训练和预测。选用序列最小优化算法和一对一的多类分类算法进行训练,建立起泥石流危险性评价的支持向量机预测模型。试验结果显示,支持向量机方法对泥石流危险性评价可以取得满意效果,具有良好的实用价值。 In view of the shortcomings of the traditional debris flow hazard assessment methods, this paper uses the support vector machine theory to train and predict the data of 10 debris flow ditch in 30 debris flow ditch in Yunnan Province and Jishishi Gorge in Yellow River. The minimum sequence optimization algorithm and one-to-one multi-class classification algorithm were used to train the support vector machine prediction model for debris flow hazard assessment. The test results show that the support vector machine method can achieve satisfactory results for debris flow hazard assessment and has good practical value.