基于遗传规划的泥石流多因子融合预测方法  

Multi-factors fusion method of debris flow prediction based on genetic programming

作　　者：翟淑花[1] 冒建[1] 南赟[1] 刘欢欢[1] 王云涛[1] 王强强[1] 熊春华[1] 王艳梅 ZHAI Shuhua;MAO Jian;NAN Yun;LIU Huanhuan;WANG Yuntao;WANG Qiangqiang;XIONG Chunhua;WANG Yanmei(Beijing Institute of Geology, Beijing 100120, China)

机构地区：[1]北京市地质研究所,北京100120

出　　处：《中国地质灾害与防治学报》2020年第6期111-115,共5页The Chinese Journal of Geological Hazard and Control

基　　金：北京市自然科学基金资助项目(8182022);北京市自然科学基金项目(8202026);北京市科委科技计划项目(Z191100001419015)。

摘　　要：泥石流是一种多发的地质灾害,常对人民生命财产安全带来极大的威胁,其暴发不仅与降雨有关,还与众多地质环境因子相关。本文以流域面积、松散物质比率、沟床平均坡度为地质因子,以最大小时雨强(T)和总降雨量(R)的乘积作为降雨指数,在获取的泥石流地质因子和降雨指数因子综合样本库的基础上,采用遗传规划法建立了泥石流临界降雨指数智能预测模型,克服了以往以雨量为单一指标的预警模型的弊端,模型验证结果显示,泥石流预测精度高、适应性强。Debris flow is a frequent geological disaster,which often poses a great threat to the safety of people’s lives and property.Outbreak of debris flow is not only related to rainfall,but also related to many geological and environmental factors.In this paper,the watershed area,ratio of loose materials and the average slope of the gully bed are taken as the geological factors,the maximum hourly rainfall intensity(T)and the total rainfall(R)are taken as the rainfall index,the sample database is established by means of geological factors and rainfall index,genetic programming is used to establish a prediction model for the critical rainfall index of debris flow,which overcomes the shortcomings of the previous warning model that used rainfall as a single indicator,model verification results show that the model has high warning accuracy and strong adaptability,which can realize timely warning.

关 键 词：泥石流 地质因子 降雨指数 遗传规划 智能识别 

分 类 号：P642.23[天文地球—工程地质学]