CAJ | 학술논문

选取南京地区滑坡为研究对象。基于DEM及遥感影像数据，选取样本，提取滑坡影响因素，计算影响因素熵值。利用K折粒子群优化（KPSO）方法，在GIS环境下针对地质资料缺乏情况，生成一个可靠的南京地区滑坡敏感图。自组织特征映射网络（SOM）法用作一个并行研究，结果作为KPSO法结果对比。结果表明，KPSO法聚类准确率为85%，自组织映射（SOM）法聚类结果准确率为80%，由此说明KPSO聚类法在地质资料不足前提下形成滑坡敏感区是一种行之有效的方法。
선취남경지구활파위연구대상。기우DEM급요감영상수거，선취양본，제취활파영향인소，계산영향인소적치。이용K절입자군우화（KPSO）방법，재GIS배경하침대지질자료결핍정황，생성일개가고적남경지구활파민감도。자조직특정영사망락（SOM）법용작일개병행연구，결과작위KPSO법결과대비。결과표명，KPSO법취류준학솔위85%，자조직영사（SOM）법취류결과준학솔위80%，유차설명KPSO취류법재지질자료불족전제하형성활파민감구시일충행지유효적방법。
In this paper, the landslide in Nanjing is selected as the study target. On the basis of DEM and remote sensing image data, landslide samples are selected for extraction of influential factors of landslide and calculation of the entropy of the factors. By use of K fold particle swarm optimization (KPSO) method, a reliable landslide susceptibility map of Nanjing in GIS environment is generated under the condition of insufficient geological data. Furthermore, self-organization feature map (SOFM) network is used as a parallel study and the study results are used for comparison with those from KPSO. The results indicate that the accuracy of KPSO clustering is 85% and that of SOFM is 80%. It can be concluded that KPSO is an effective method for landslide susceptibility analysis under the precondition of insufficient geological information.