摘要
目的:为减少滑坡带来的损失,本文提出了一种优化后的支持向量机(SVM)模型来进行滑坡预测。方法:通过模拟实验获得多属性传感器数据,对数据进行熵值法处理,获得分类标签,再通过灰狼算法(GWO)优化后的支持向量机模型进行预警。结果:熵值法划分后的数据,通过GWO-SVM模型进行建模预测,得到的预警结果与实际结果高度吻合,预测精度在95%以上。结论:熵值法能处理滑坡过程中的多属性数据,适用于滑坡的复杂非线性情况。对比单一支持向量机模型,优化后的模型精度更高,稳定性更好,能对当前滑坡易发性进行精确地预测。
Aims:An optimized support vector machine(SVM)model of landslide prediction was proposed to reduce the loss caused by landslides.Methods:Multi-attribute sensor data was obtained through simulation experiments;and the data was processed by the entropy method to obtain classification labels.Then the support vector machine model was optimized by the gray wolf algorithm(GWO)for early warning.Results:The GWO-SVM model was established with data divided by the entropy method,with a prediction accuracy of over 95%.Conclusions:The entropy method can process the multi-attribute data in the landslide process,which is suitable for the complex nonlinear situation of the landslide.Compared with the single support vector machine model,the optimized model has higher accuracy and better stability,and can accurately predict the current landslide susceptibility.
作者
田文财
李青
TIAN Wencai;LI Qing(College of Mechanical and Electrical Engineering,China Jiliang University,Hangzhou 310018,China)
出处
《中国计量大学学报》
2021年第2期253-259,共7页
Journal of China University of Metrology
基金
国家重点研发计划课题(No.2017YFC0804604)
国家质量监督检验检疫总局科技计划项目(No.2017QK053)
浙江省重点研发计划项目(No.2018C03040)。
关键词
滑坡
熵值法
灰狼算法
支持向量机
非线性
landslide
entropy method
gray wolf algorithm
support vector machine
nonlinear
