基于随机森林的洪灾风险评价模型及其应用

发布时间：2018-06-26 21:37

本文选题：洪灾风险 + 随机森林　；参考：《水利学报》2015年01期

【摘要】：依据流域灾害系统理论,在综合考虑致灾因子、孕灾环境和承灾体的基础上选取10个评价指标,构建了基于随机森林(RF)智能算法的洪灾风险评价模型,借助GIS技术以100 m×100 m栅格为计算单元,对东江流域进行了评价。结果表明:该模型设置参数少,且无需设定指标权重和分级标准,实现过程较简单;RF模型拥有指标重要性评估功能,方便分析各评价指标对洪灾风险的贡献度;模型的评判精度、分级结果均比支持向量机(SVM)的结果更好,数据挖掘能力更强;与GIS技术结合,便于分析洪灾风险的空间格局及内在规律,具有较好的适用性,为流域洪灾风险评价提供了新的思路。
[Abstract]:According to the theory of watershed disaster system, a flood risk assessment model based on the intelligent algorithm of random forest (RF) was constructed by selecting 10 evaluation indexes on the basis of comprehensive consideration of disaster factors, disaster environment and disaster bearing body. Taking 100m 脳 100m grid as calculation unit, the Dongjiang river basin was evaluated by GIS technology. The results show that the model has less parameters and does not need to set the index weight and classification standard. The RF model has the function of evaluating the importance of the index, and it is convenient to analyze the contribution of each evaluation index to flood risk. The evaluation accuracy and classification result of the model are better than those of support vector machine (SVM), and the ability of data mining is stronger. Combining with GIS technology, it is convenient to analyze the spatial pattern and inherent law of flood risk, and has better applicability. It provides a new idea for flood risk assessment.
【作者单位】：中山大学水资源与环境研究中心;华南理工大学土木与交通学院;华南地区水循环与水安全广东普通高校重点实验室;
【基金】：国家自然科学基金资助项目(51209095,51210013,51479216) 国家科技支撑计划(2012BAC21B0103) 水利部公益项目(201301002-02,201301071) 中央高校基本科研业务费专项基金(2014ZZ0027)
【分类号】：TV87;X43

【相似文献】