Reservoir Bank Collapse Prediction Of Fengdu County In Reservoir Area Of Three Gorges Based On Support Vector Machine And Extenics

The Yangtze river three gorges reservoir has began to three stage impoundmentsince June,2003, through135m and156m and175m water storage process, thewater level of dam front running between145m and175m. The three gorges waterproject has appeared good benefit in power generation, shipping, water supply,drought, flood control, ecological aspects. Three gorges reservoir area which has thecomplex geological conditions is located in the west of Chongqing and the east ofHubei, and vulnerable to the impact of factors such as the rain, has always beengeological disasters multiple areas such as landslide, collapse, debris flow. Since2003water storage, reservoir geological disasters such as landslide activities tend tofrequent, disaster. The Yangtze river three gorges reservoir is mountain stream typereservoir, the total length of the reservoir bank about5300km,441km of which haspoor reservoir Bank stability. After the impoundment of the three gorges reservoir,area between145m and175m of the reservoir bank within the scope of naturalcondition change significantly, resulting in the base level of erosion and undergroundwater level up, the library water soak time variable length, fluctuation amplitudeincreases, strengthening groundwater dynamic effect, and speeding up the libraryshore bank collapse process, causing serious influence of the reservoir bank people’slife and property and production life, ecological environment, reservoir operation,bank collapse is a major geological disasters problem of influencing the ThreeGorges Reservoir Area.Reservoir bank collapse research results at home and abroad more focus on thesoil bank primarily plain region impact abrasion type library shore reservoir bankcollapse problem, for three gorges reservoir area this kind of mountain stream typereservoir bank collapse having relatively insufficient research, it is urgent toestablish suitable for the three gorges reservoir area of the reservoir bank stabilityevaluation and bank collapse prediction model and method system. Engineering andtechnical personnel and scholars have widely carried out in three gorges river typereservoir characteristics of stability evaluation of the reservoir bank, from differentperspectives putting forward the method of bank collapse prediction, but the study generally are in exploration stage, still need further research and practiceverification.The paper is based on taking bank geological conditions survey projects of thethree gorges reservoir area, through the three gorges reservoir of Chong Qing sectionof the site geological conditions investigation, summarizes the Fengdu Countynatural geography and hydrogeology condition, to Fengdu County Yangtze rivermain stream and main tributaries of the engineering geological conditions of thereservoir bank on detailed investigation and research, on the influence of the mainfactors of the slope stability research, combining both research results and the fieldgeology survey material, studying influencing reservoir bank collapse failure modeof Fengdu County, the bank collapse prediction parameter quantitative evaluationand value, based on the nonlinear theory method to establish reservoir bank collapsenonlinear evaluation prediction model and method system for the three gorgesreservoir geological disaster prevention project to provide basis.The paper set up for the first time basing on kernel principal component-leastsquares support vector product-extenics nonlinear reservoir bank caving evaluationprediction model, taking Fengdu County as the research object of rock slope, to the156m and175m of reservoir bank collapse evaluating prediction research. Thepaper is based on kernel function theoretical research, using the ability of Kernelfunction strong nonlinear data processing and mining, to the reservoir bank collapseof the influence factors to the dimension reduction processing. In the Matlabsoftware platform, to the Fengdu County bank collapse of the main influence factorsto the extraction basing on kernel function of principal component analysisalgorithm, and has obtained the good effect. Based on least squares support vectormachine and extenics bank collapse prediction research lay the foundation. Thethesis on least squares support vector machine theory, many kinds of classificationproblem, and the selection of kernel function and optimization has done research. Inthe Matlab platform construction of the reservoir bank collapse prediction leastsquares support vector machine model, adopt “one-to-many method” classificationmethod respectively on156m and175m of reservoir bank stability prediction, andthe predicted results and experience formula calculation results are compared, andachieved good results. On the basis of the theory of extenics, this paper studies theextension evaluation modeling process and method, the weight coefficient ofcorrection, realize a good evaluation effect. Using the software to finish my paper the relevant achievements operation,after back calculation test, the programcalculated results of prediction results correctly, and the program algorithm isreasonable.Taking reservoir bank slope rock of Fengdu County as an example, theapplication of the thesis established reservoir bank caving nonlinear predictionmodel of156m and175m storage level bank stability level evaluation predictionand forecasted the possible scope of the influence of bank collapse.Using thegeographic information system and visualization techniques, proceeds the reservoirbank for3D visualization and qualitative auxiliary analysis. By comparison, analysis,research, it is known that based on nonlinear evaluation results is more accordingwith reservoir shore practical than the traditional method, and more objective andreasonable. Research results for the three gorges reservoir Fengdu County and otherreservoir bank collapse prediction and prevention and control project of the reservoirbank provide reference for researchers.Through reservoir bank collapse prediction research of the three gorgesreservoir area of the reservoir bank Fengdu County, obtained the following mainresults and conclusions:1.By using nonlinear theory method firstly sets up basing on kernel principalcomponent-least squares support vector machine-extenics nonlinear reservoir bankcaving forecast model. This model can be used in different structure types, differentbank collapse mode, different storage level bank of reservoir bank stabilityevaluation and bank collapse prediction, model can be widely used in mountainstream type reservoir bank collapse prediction, good portability.2.Using kernel principal component theory to Fengdu County reservoir bankcaving main influence factors to the extraction, In the influence height and width ofthe11influence factors and the factors influencing the stability of the reservoir bank13factors, extracts respectively7and9factors of the strongest related principalcomponent, the calculation results kernel principal component replace the originalinfluence factors evaluating and predicting3.Introducing for the first time based on least squares support vector machinemethod proceed reservoir bank caving and influencing height, width prediction.For14coastal library of Fengdu County, respectively the156m and175m storagelevel conditions, respectively by means of least squares support vector machine and extenics two nonlinear method proceeding reservoir bank collapse evaluation and theinfluence range prediction calculation.4.By comparing least squares support vector machine and extension theory thattwo methods of bank collapse prediction results analyzing that two kinds ofnonlinear reservoir bank collapse prediction results are basically similar, and withfield investigation conclusion consistent, Verify the ability of the nonlinear theoryprocessing reservoir bank collapse question, explaining that the adoption ofnonlinear theory constructing mountain stream type reservoir bank collapseprediction model is scientific and feasible, the study method is innovative, has theapplication and popularization of value, embodies the thesis and research withscientific significance and practical value.5.Based on the ARCGIS geographic information platform, introducesgeographic information visualization techniques, for the landslide of the reservoirbank proceeding3d visualization, auxiliary analysis. The geographic informationsystem platform of the reservoir bank of3D visualization increases intuitiveness ofthe stability evaluation of the reservoir, and for bank collapse height, widthprediction quantitative prediction with the ability to rise.