泥石流危险性SIGA-BP神经网络评价方法及应用  

Method and Application of Debris Flow Hazard Assessment Based on SIGA-BP Neural Network

作　　者：谷秀芝[1] 陈洪凯[1] 刘厚成[1] 

机构地区：[1]重庆交通大学岩土工程研究所,重庆400074

出　　处：《重庆交通大学学报（自然科学版）》2010年第1期98-102,共5页Journal of Chongqing Jiaotong University(Natural Science)

基　　金：国家自然科学基金项目(50678182);中国博士后科学基金项目(20080430095)

摘　　要：泥石流危险度是由泥石流危险因子综合判定的,然而危险因子有主次之分,要从众多泥石流危险因子中筛选出作用最大的主要危险因子是很困难的,利用自适应免疫遗传算法SIGA(Self-adaptive Immune Genetic Algorithm)对BP神经网络进行优化,获得了与云南省最相关的7项泥石流危险因子,建立了基于SIGA的BP神经网络模型,并对10组泥石流沟数据进行预测,得到了较高的预测结果。The risk degree of debris flow is determined by dangerous factors of the debris flow. The dangerous factors are di- vided into primary and secondary factors. It is difficult to choose the most dangerous factor. BP neural network is optimalized by self-adaptive immune genetic algorithm （ SIGA）, and seven dangerous factors of Yunnan province are obtained. SIGA-BP neural network is also established, which is applied to forecasting data of 10 groups＇ debris flow, and more accurate fore- casting results are obtained.

关 键 词：泥石流 神经网络 危险度 危险因子 

分 类 号：P642.23[天文地球—工程地质学]