Research On Multi-models Comprehensive Forecasting Of Landslides Based On Fuzzy Integral Method

Landslide disaster has always beenthe largest number of most serious harm toone of the types of geological disaster. It not only causes soil erosion, environmentaldestruction, but brings heavy casualties and property losses. Landslide disaster occursmainly in gully aspect area, rich rainfall area.China is one of countries wherelandslide disaster occurs frequently and seriously. Currently,the most effectivemeasures for reducing landslide losses are extensive landslide monitoring andforecasting, and improving forecast accuracy.In this paper,firstly,Huangci landslide, Saleshan landslide and other landslideexamples are chosen to analyze forming mechanism from theirs type, lithology,geology, hydrogeological conditions and the stages they were in.At the same time,summarize reduction methods based on the creep and their applicability. Threeprediction models-Verhulst model, Verhulst inverse-function model and the f methodof landslide time prediction are chose to analyze and forecast the above landslideexamples. Results ofanalysisshow that the same forecast modelhas largedifferencesfor different types of landslide; the different modelsimilarly has largedifferences for the same landslide, therefore the single forecast model has itslimitations.Multi-source information fusion method is summarized.The fuzzy integralmethod and D-S evidence theory are chosen according to respectivecharacteristicsand advantages.The two fusion methods areboth based on themembership degree. Combine the expert knowledge with the othernessbetweenaheadand retardation time of the singleprediction model and design the membershipfunctionsin line withpractical situation to blur the prediction results of the singlemodel. Utilizing two fusion methodsselected to calculate the result of fusion. Fusionresults show that:(1) the result through two kinds of fusion methodsis improvedagainst single model;(2) to the two fusion methods, fuzzy integral’s accuracy ishigher than D-S evidence theory;(3) the fuzzy integral method has applicability incomprehensive prediction of landslide.Multi-source information fusion method can effectivelycoalesce advantages ofeach single model to improve the comprehensive prediction accuracy of forecastingobjectby certain rules.Fuzzy integral method of multi-source information fusion usesthe Intersect and the Union of fuzzy mathematics, and reduces the computationalcomplexity. This method is more convenient.