基于数据驱动模型的区域滑坡危险性评价模型综述  

A REVIEW OF REGIONAL LANDSLIDE HAZARD ASSESSMENT MODELS BASED ON DATA-DRIVEN MODELS

作　　者：张柯月 崔玉龙[1] 钱志冲 ZHANG Ke-yue;CHUI Yu-long;QIAN Zhi-chong(School of Civil Engineering and Architecture,Anhui University of Science and Technology,Huainan 232001)

机构地区：[1]安徽理工大学土木建筑学院,安徽淮南232001

出　　处：《云南地质》2024年第1期121-127,共7页Yunnan Geology

基　　金：国家自然科学基金项目(42277136)。

摘　　要：滑坡是最常见的地质灾害之一,具突发性、随机性、范围广、危害大等特点。本文总结了基于数据驱动区域滑坡危险性评价模型,分析数理统计中信息量模型、逻辑回归模型、粗糙集理论、证据权模型、频率比模型、熵指数模型,机器学习方法中支持向量机、决策树、随机森林、XGboost、Catboost、神经网络模型,深度学习方法中卷积神经网络等模型在应用中优点和局限性,对数据驱动模型区域滑坡预测模型存在的部分问题分析与讨论。As one of the most common geological disasters,landslide has the characteristics of sudden,random,wide range and great harm.In this study,the advantages and limitations of the application of different models were summarized,including the data-driven regional landslide risk assessment model,models of mathematical statistics method such as information model,logistic regression model,rough set theory,evidence weight model,frequency ratio model,and entropy index model,models of machine learning method such as support vector machine,decision tree,random forest,XGboost,Catboost and neural network,and convolutional neural network model of deep learning method.Based on this,some problems existing in the data-driven regional landslide prediction model were analyzed and discussed.

关 键 词：数据驱动 模型综述 危险性评价 滑坡 

分 类 号：P642.21[天文地球—工程地质学]