Landslide Susceptibility Evaluation Based On Support Vector Machine Model And Evidence Theory

With the rapid socio-economic growth,the Badong and Zigui sections of the Three Gorges reservoir area are affected by many factors such as complex geography,variable geological structure and stratum lithology,and frequent landslides and other disasters.Therefore,it is of great value to carry out research on landslides.Using support vector machine model and evidence theory,a detailed application was carried out in Padong and Zigui District,and the scientificity of the method was verified.The main work and results of the paper are as follows:(1)In-depth study on the attributes,characteristics,lithologic grouping,and multi-value extraction of sample data involved in landslide susceptibility evaluation.Three evaluation factors(three lithologic factors and seven environmental impact factors)were extracted from Padang to Zigui section of the Three Gorges of the Yangtze River.(2)Uncertainty analysis was performed on the model.Ten different data sets were randomly selected from the training sample data.The PSO-SVM coupling model was used to train and evaluate these 10 data sets respectively,and the final results were obtained.The ROC curve analyzes the accuracy of the assessment results based on 10 different data sets and the PSO-SVM coupling model for the study area as high as 86% and very close.At the same time,the PSO-SVM model is consistent with the evaluation results of different data sets in terms of high susceptibility and low susceptibility.(3)Through the integration of uncertainty of evaluation factors,the total area of the study area is divided into four categories: low,medium,high,and extremely high according to 40%,30%,20%,and 10%.The PSO-SVM model is compared with the ROC curve evaluation results of evidence theory.It is found that the PSO-SVM coupling model is feasible and superior to the evidence theory.(4)The basic probabilistic distribution function is used to define the basic probability distribution functions of all evaluation factors,and the performance law of the trust function on the evaluation factors such as elevation,slope aspect,slope type,drainage area,and water system height is studied as follows: Large and decreasing,when the distance from the water system is 1650-2650 m and the lithology of the rock group is represented by soft rock,the values of the trust function are 0.273 and 0.271 respectively.