| With the development and progress of human society, people’s demand for resourcesgrows rapidly. In today’s society, the energy problem has become an important factorrestricting the economic development; the energy crisis is becoming more and moresignificant. In this context, water resources, as a kind of clean and renewable energy,receive more and more attention all over the world. Our country has abundant waterresources, the amount that can be developed ranks first in the world, but thedistribution is extremely uneven, most concentrated in the southwest region. At thesame time, the southwest region is located in the east of Qinghai Tibet plateau, theunique regional geological background lead to a harsh nature condition, complexgeological environmental condition and the frequent occurrence of nature disasters. Inaddition to more and more frequent human activities in recent years, the balancesituation of large-scale construction has changed the existing geological conditions,triggering a new geological disaster. Among them, the landslide disaster has thecharacteristics of sudden, complex factors and high hazard, which poses a great threatto the life and property safety of the people within the scope of the hydropowerprojects. Therefore, developing the landslide stability evaluation and landslide hazardrisk analysis research, provide scientific basis for the landslide control, disastersprevention and reduction and the risk management. And it is of great academic valueand practical significance to ensure the rational development and utilization of resources and protect the environment for human survival.In this paper, take Benzilan Reservoir area of Yunnan diversion project as thestudy area, with24typical landslides in the region and development as the mainobject of study, on the basis of collecting regional geological data, combining on theintegrated use of interior remote sensing interpretation and field geological surveymethod to analyze the basic characteristics of the landslide, distribution in space anddeformation mechanism, select single landslide to evaluate stability indeices andfeature extraction, using modern computational intelligence methods to assess thestability of landslide, the establishment of regional landslide hazard evaluation systemto determine the weight of each index, the study area for landslide hazard zonationanalysis, choose landslide hazard vulnerability factor and vulnerability analysis, thefinal take division of the study area by the risk analysis. Based on the above research,the following conclusions are drawn:1. The reservoir area landslide mainly concentrated in two places: Tail Darizone and Guxue rural area along the upper reaches of the Dingqu River. According tostatistics of the results of the scale landslide, rock structures and the causes ofdeformation mechanisms show that: Reservoir landslide showing in a thick layer thangreat thick layer, large layer than giant layer, and bending-pull-based than sliding-bending genetic type.2. Selecting11single landslide stability evaluation indices, applicatingevaluation data FastICA algorithm for feature extraction, eliminating the correlationbetween various data evaluation, acquisition independent components which reflectthe nature information of landslide stability, thus it is more effective in identifying thedifferent types of landslide stabilization mode.3. The extreme learning machine intelligent evaluation model of monomerlandslide stability was established by using the extreme learning machine theory. Theresults of the stability analysis of landslide in reservoir area showed that there are2landslide samples in the stable state and the proportion is about16.67%of all12landslide samples used for prediction,5landslide samples are in the basic stable stateand the proportion is about41.66%,2landslide samples are in the potential unstable state and the proportion is about41.66%, and there are3landslide samples in theunstable state and the proportion is about25%.4.9evaluating indicators of regional landslide hazard risk were selected, thesubjective weight of each evaluating indicator are determined based on the weightedleast-square method and DEMATE method, and objective weight of each evaluatingindicator are determined based on entropy method and CRITIC method. At last, thecombination weight is obtained by the combination of the subjective weight andobjective weight using distance function. Weight analysis results show that: themaximum weight is terrain slope, the secondary is fluvial erosion and the minimumweight is geological structure.5. The reservoir landslide disaster risk division is based on clustering model andinformation model, respectively. The division results show that both clustering modeland information model can reflect the characteristics which the high risk of landslidedisaster zone distributed along the Jinsha River and Dingqu River. Considering theaccuracy of the division, the accuracy of the clustering model is slightly higher thanthe information model. And the division result of clustering model show that there arealtogether32.72%areas at high risk and very high risk area.6. The results of landslide disaster vulnerability division show that, high and veryhigh vulnerability areas are mainly located in the village where human activities arefrequent and life and living facilities are densely covered. The results of landslidedisaster risk division show that there are altogether25.92%areas at high risk and veryhigh risk area, mainly distributed in the area where the landslide disaster are moredevelopment along Jinsha River and Dingqu River and the production and life ofhuman beings are more concentrated. |
PDF Full Text Request