Early Intelligence Indentificaition Methods Of Stability And Fuzzy Synthetic Prediction Of Hazard For Landslides In Rimian I Lydropower Station Reservoir

Along with the increasingly tight energy crisis, the demand of energy forcountry’s development is much more growing. Hydropower resources, as clean andcheap renewable energy, is coused a great attention. In southwest China, there are alot of water resources, because river systems are distributing, landform type ismountain, river divide is great, and rainfall is relatively concentrated. The landslidesoccur quickly, large quantities, high hazardous, causing great harm to human life andproperty of the hydropower station reservoir area. More and more research scholars athome and abroad take a great deal of attention on the landslide geological disasters.The academic thought of this paper is “mechanism analysis of geologicalprocess” and “the quantitative appraisal”. The3S technology with traditionalgeological surver methods and laboratory analysis with field surver instruments,basing remote sensing images and three-dimensional digital model, are introduced foranalyzing the stability of reservoir bank and landslides, and landslide hazard in thehydropower station reservoir area. The research provides a scientific basis forhydropower project implementation.There are two locations with concentration of landslides distribution in theRimian hydropower station reservoir, after researching the spatial distributioncharacteristics of landslides along the reservoir area. The first location is near theXulong dam site along the Jinshajiang River. The Xulong dam site has not eventually been selected. And the second lacation is near the tail of reservoir along Wangdalongvillage. The research result shows that geological stratum and structure play a greatimportant role at landslide formation. Most landslides form and take place at DTJstratum and structural development area.There are many factors affecting the stability of reservoir bank, such asstratigraphy, structure, topography, hydrology, other predisposing factors, and so on.The stability characteristics of reservoir banks show that there are29bank coastsalong two banks of Jinshajiang River in the reservoir area. There are8unstable bankcoasts,16stable and basically stable coasts, and5extremely stable bank coasts. Mostof landslides are in unstable bank coasts.It is necessary to reduce and simplity the factor attribute. The factor contributionrate analysis method and rough set theory are used for factor attribute reduction. Atlast, the number of choosen factors is8. They are rockmass structure, slope structure,the height of the slope, average slope, deformation signs, fault distance, the floodedproportion, induced seismicity.The intelligent model for landslides early stability in the reservoir area isestablished by support vector machine theory (SVM) which is improved by posteriorprobability analysis. The stability characteristics of landslides in Rimian hydropowerstation reservoir area are evaluated by SVM model. The result shows that, there are10stable landslides accounting for45.45%,8basically stable landslides accounting for36.36%,3potentially unstable landslides accounting for13.64%,1unstable landslideaccounting for4.55%.In this paper,8factors are extracted to evaluate the hazard of landslides. Theyare landslide stability characteristics, sliding velocity, sliding distance, seell height,distance from the dam site, population of sliding risk range, road distance and landresource type. The weights of these factors are computed by subjective and objectivemethods. The weighted least square method is used as subjective method decided byspecialist experience, and the entropy theoty method is used as subjective methoddecided by factor data characteristics. The result weights of every factor are definedby combined weight process basing on the two methods above, using distribution coefficients computed by distance function. The finally composite weights of8factors are0.1901,0.1856,0.1292,0.0954,0.1264,0.1446,0.0702and0.0584.Fuzzy comprehensive method is using to evaluating the hazard degree oflandslides in the station reservoir area. The membership function of each factor isestablished in the paper. The hazard characteristic of every landslide is evaluatedbasing on the maximum membership principle. The result shows that, there are5extreme hazard landslides accounting for33.33%,4moderate hazard landslidesaccounting for26.67%,6mild hazard landslides accounting for40.00%.