Research And Application Of Uncertain Genetic Ant Clustering Algorithm Based On Approximate Backbone In Landslide Hazard Prediction

Landslide hazard is an important threat that affects the life safety and natural environment of our country at present.There are many uncertain factors affecting the landslide occurrence which make it so hard to predict the landslide hazard.Therefore,finding an effective way to predict landslides is an urgent issue to solve landslide disaster.The landslide are affected by multiple factors which have different complexity and correlation and have a guiding role in landslide risk prediction.In order to deal with the relationship effectively between influence factors and landslide hazard,we use ant colony clustering algorithm to the data objects of high similarity to divide a class,introduce genetic algorithm and update the pheromone rule to optimize the ant colony clustering algorithm which is easy to fall into local optimum problem.It uses the theory of large scale data processing framework approximation to solve the problem of stagnation and slow convergence of the traditional ant colony clustering algorithm combined with genetic algorithm.The approximate skeleton theory is constructing the approximate skeleton genetic ant colony clustering algorithm model,and integrating the landslide disaster-related characteristics,designing landslide hazard prediction models,and studying the effectiveness of the algorithm in landslide risk prediction experiments.However,rainfall is a continuous value and the approximate skeleton genetic ant colony clustering algorithm can't accurately describe it because of its uncertainty in landslide risk prediction.In order to solve the difficulty of characterizing the uncertainty of rainfall effectively,Combining Gauss interval value data processing method and point probability model,an uncertain data processing model is designed to characterize the uncertainties such as rainfall.The synthetic of the approximate backbone theory,the uncertain data processing model and the improved ant colony clustering algorithm are used to construct the uncertain approximate backbone ant colony clustering algorithm.An uncertain UCI data experiment is designed to verify whether this algorithm has high validity.Finally,the algorithm is applied to the experiment of Yan'an Baota landslide in Shaanxi province.We build the model of landslide hazard prediction and verify the possibilities of the algorithm in landslide hazard prediction.