| Since the founding of the People's Republic of China,a large number of rich water resources and hydropower resources have been developed in the middle and upper reaches of the Yangtze river.Because of the complex topography and steep slopes of the Yangtze river basin,the study of these landslides has cost a lot of manpower and resources.Artificial intelligence algorithm is an intelligent computing model that simulates the thinking mode of human beings.It has been applied to various engineering fields and solved many complex engineering problems.Early in this article,the author based on the water by controlling the phase down water drop rate increase with small deformation wading research on deformation control mechanism of the landslide stability,the deposit ceiling field dam hydropower station village and study of monitoring and early warning and handling measures,as well as the large reservoirs in the run-time accumulation landslide self-organizing adaptive prediction model of deformation and deformation mechanism "project,hydropower station reservoir area in three gorges reservoir area and ceilings,much work has been carried out,the reservoir induced accumulation of landslides have a more thorough understanding.On the basis of artificial intelligence will be the new technology combined with traditional landslide deformation theory,by means of several relatively mature artificial intelligence algorithms,with accumulation of landslide deformation mechanism and deformation response law as the foundation,set up based on artificial intelligence algorithm library induced water accumulation landslide deformation prediction model,in order to improve the reservoir water induced the accumulation of the landslide deformation prediction research in the field of work efficiency,and mainly do the following work:(1)abundant geological data and monitoring data have been obtained through the field investigation and data collection of a large number of landslides in the three gorges reservoir area,such as the babimen landslide,the baijiabao landslide,the sanmendong landslide,the niuli landslide,and the tianba village deposit landslide in the reservoir area of the ceiling hydropower station.(2)based on the summary and analysis of landslide geological data and monitoring data,combined with the latest achievements in related fields,a method to identify the landslide deformation pattern of reservoir water induced accumulation layer is proposed,which takes the geological structure of the landslide and the meteorological and hydrological conditions as the characteristic points.By summarizing and analyzing the geological structure characteristics,deformation response law and monitoring data curve of the landslide,the landslide is divided into ladder type,trend type,random type and continuous growth type according to the deformation mode.Based on matlab,LIBSVM toolbox is used to identify and classify landslides.The basic geological and hydrological data such as slope length,slope width,volume,gravel content,permeability,submerged ratio of reservoir water level were coded,and the sample landslide was identified and classified.To some extent,it is explained that the deformation of the landslide is the result of the comprehensive influence of geological structure and hydrometeorological conditions.(3)BP neural network,wavelet neural network(WNN),extreme learning machine(ELM),genetic algorithm(GA),particle swarm optimization(PSO)and other more mature algorithm models in the field of artificial intelligence were selected,and the calculation model was selected according to each deformation mode of the landslide.The wavelet function is used to decompose the original monitoring data into the trend term data which is easy to predict and the period term data which is more complex.Genetic algorithm and particle swarm optimization algorithm were used to optimize the BP neural network,and weighted average was used to calculate the results.It has a good prediction effect for the period term data of ladder type landslide.Basically conforms to the deformation law.Wavelet neural network is used to predict the deformation of random oscillating landslide.By studying the response relationship between the changes of rainfall and reservoir water level and the landslide deformation,the parameter weight of 3:7 is set,and it reflects the response rule of the real deformation,reservoir water level and rainfall.The same method is also applied to the period term data of trend oscillating landslide.(4)taking tianba village as an engineering example for verification and analysis.Through analysis and demonstration,the landslide is divided into two deformation stages with January 2016 as the boundary,each belonging to a different deformation mode.The latest geological data and hydrometeorological data of the landslide were collected,and the deformation mode of the landslide in the tianba village accumulation layer was accurately identified by using LIBSVM toolbox.It is proved that the deformation pattern of landslide is influenced by geological structure and hydrological conditions.RMSE is 1.34.(5)artificial simulation of the tianba village accumulation layer landslide in two different conditions.By changing the water level and meteorological conditions,the influence of sudden fall of reservoir water level and heavy rain on landslide can be predicted in half a year.When the predicted results are under normal conditions,the maximum deformation of the landslide in the next half year reaches 53.9mm,which is 3.8mm higher than that of half a year ago.Under the condition of sudden drop of reservoir water level and heavy rain,the total displacement of the landslide reached 78.9mm after half a year,which was 28.8mm higher than that of half a year ago.It can be seen that the landslide will remain stable without overall instability even after the reservoir water level plummeting and the extreme conditions of rainstorm.This is also consistent with the conclusion in the engineering research report,which can provide reference for the evaluation and research work of the landslide. |
PDF Full Text Request