| 基于LR-ANN-SVM的滑坡易发性评价 |
| |
| 引用本文： | 陈飞,蔡超,李小双,钱乾.基于LR-ANN-SVM的滑坡易发性评价[J].有色金属科学与工程,2020,11(4):82-90. |
| |
| 作者姓名： | 陈飞 蔡超 李小双 钱乾 |
| |
| 作者单位： | a.江西理工大学，资源与环境工程学院，江西赣州 341000 |
| |
| 基金项目： | 国家自然科学基金;江西省教育厅科技项目;赣州市专利技术实施转化资助项目 |
| |
| 摘 要： | 针对传统大数据机器学习等方法进行滑坡易发性评价时，存在过于追求模型评价精度，导致在中易发区与低易发区存在滑坡产生的风险，提出了风险预警来降低中与低易发区产生的滑坡灾害。选取神经网络模型（ANN）、逻辑回归模型（LR）、支持向量机模型（SVM）3种学习方法，对上犹县进行滑坡易发性评价，将上犹县分为高易发区、较高易发区、中易发区、较低易发区，低易发区。由受试者工作曲线（ROC）下的面积（AUC)显示:神经网络（ANN）的AUC=0.939, 逻辑回归模型（LR）的AUC=0.897, 支持向量机（SVM）的AUC=0.884，均具有较高的评价精度。根据以上的易发性评价结果，得到上犹县栅格的易发性指数（LSI），然后基于MAX（LSI（LR）、LSI（ANN）、LSI（SVM））函数对上述模型的易发性指数取最大值，并对上犹县进行滑坡易发性评价。结果显示:LR-ANN-SVM的AUC=0.815，有较高的易发性评价精度。从高易发区与较高易发区所含滑坡占比来看，LR、ANN、SVM、LR-ANN-SVM的滑坡占比分别为80.6%、74.6%、91%、93.2%，表明根据ANN-LR-SVM易发性分区治理更安全。
|
| 关 键 词： | 神经网络 支持向量机 逻辑回归模型 滑坡风险预警 |
| 收稿时间： | 2020-05-18 |
Evaluation of landslide susceptibility based on LR-ANN-SVM |
| |
| Affiliation: | a.School of Resource and Environmental Engineeringb.Jiangxi Key Laboratory of Mining Engineering, Jiangxi University of Science and Technology, Ganzhou 341000, Jiangxi, China |
| |
| Abstract: | In the susceptibility analysis for landslide, methods like traditional Big Data machine learning are over-emphasis on evaluation of the accuracy of the model. Landslides risk warning will be given toreduce damages in medium-susceptibility and low-susceptibility areas. Three common learning methods-artificial neural network (ANN), logistic regression (LR), support vector machines (SVM) -- were selected in this research to evaluate landslide susceptibility in Shangyou County. Shangyou County was divided into high, higher, medium, lower, and low susceptibility areas. Shown by the values of the area under the curve (AUC): AUC of artificial neural network (ANN)=0.939, AUC of logistic regression (LR)=0.897, AUC of support vector machine (SVM)=0.884. The data have high evaluation precision. According to the above evaluation, the latent semantic index (LSI) of the raster in Shangyou County is obtained. Based on the MAX (LSI (LR), LSI(ANN), LSI(SVM)) function, maximum value of thesusceptibility of the above model was obtainedto evaluate the susceptibility of Shangyou County. The results show that the AUC of LR-ANN-SVM=0.815, which has a relevantly high accuracy of susceptibility evaluation. According to the proportion of landslides in the high- susceptibility areas and the higher- susceptibility areas, the proportions of landslides in LR, ANN, SVM, and LR-ANN-SVM are 80.6%, 74.6%, 91%, and 93.2% respectively, indicating that ANN-LR-SVM susceptibility partition governance is more secure. |
| |
| Keywords: | |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
| 点击此处可从《有色金属科学与工程》浏览原始摘要信息 |
|
点击此处可从《有色金属科学与工程》下载全文 |
|
