Research On Landslide Geological Hazard Prediction Model And Application

There are many vast mountainous areas in China that have been restored,and landslides often occur during the rainy season.The occurrence of landslides can bring huge losses to human society.Relevant departments have been committed to researching how to use existing technologies to reduce the losses caused by landslides.With the development of big data and other technologies,constructing landslide prediction models to predict the risk and time of landslide occurrence has become the most favored method by many researchers.This article takes landslides in the Sichuan Tibet Railway area as the research object,and will conduct research from the following three aspects.Firstly,using image recognition technology to identify areas prone to landslides;Secondly,construct a landslide geological disaster time prediction model and a risk prediction model to predict the time and risk of landslide occurrence;Finally,develop an online landslide geological hazard prediction system for visualizing landslide prediction results.The final results of this article are as follows:(1)A optimization method is proposed to address the shortcomings of existing techniques for calculating three-dimensional geometric variables,and the calculation results are compared with sensor monitoring results.The final comparison result shows that there is very little difference between the two,so this method can be used for calculating surface deformation variables.(2)In response to the existing use of manual exploration to identify landslide prone areas,an optimized Faster RCNN algorithm is proposed to construct landslide prone areas.The experimental results show that the prediction accuracy of the model is89%.(3)Aiming at the problem that the accuracy of surface shape variable prediction is not high enough,a method based on fixed factors and induced factors to predict surface shape variable is proposed.This method first uses the moving average method to decompose the surface shape variables into fixed factor items and induced factor items,then uses the polynomial regression algorithm(PR)to predict the fixed factor items,uses the improved BP neural network to predict the induced factor items,and finally forms the surface shape variable prediction model.The model improves the prediction accuracy of the surface variables.(4)In view of the low prediction accuracy of the existing landslide geological hazard risk prediction model,the classification landslide prediction is innovatively proposed,and the Apriori association rule mining algorithm is used to mine the main influencing factors of different types of landslides,so as to build the training set of different types of landslides.Finally,the least squares vector machine(LS-SVM)algorithm is used to train the landslide geological hazard risk prediction model.(5)Develop the landslide warning system,build the OpenTSDB time series database cluster to solve the storage of a large amount of stream data,use the Redis database to cache the data to improve the response time of the system,and use the multi-thread programming method to realize the risk prediction of multiple monitoring points in parallel.