Health Evaluation And Prediction Of Slope Based On Intelligent Methods

With the advancement of science and technology and the development of productive forces,we know that the health of the structure is extremely important.Once there is something wrong with it,not only will it cause significant losses of economic,but also it will cause casualties and even more serious accidents,the consequences would be disastrous.Therefore,it is very significant to evaluate and predict the health status of the structure.Initially,health evaluation is defined in terms of people's health.Health evaluation refers to the evaluation of the health degradation of a system.According to the monitoring information of the system,we can obtain the conclusion of the system's fault diagnosis as well as the level of confidence.What's more,according to the history information,operating status and operational load characteristics of the system,we can predict the future health status of it.In order to verify that the method of the health evaluation and prediction is feasibility,in this thesis we regard the slope as the subject.Taking the characteristics of the slope itself into account,we know the health state of the slope is mainly reflected by its stability,so we can attribute the study of the health state to the study of the stability of the slope.Slope stability refers to the stability of the slope rock or soil at a certain slope height and slope angle.The main contents of this thesis are as follows:Firstly,this thesis studies the methods of health evaluation are adopted in different fields.In the light of the characteristics of the slope,it establishes health evaluation index system which is suitable for the slope.The improved analytic hierarchy process is combined with the fuzzy mathematics theory,the improved AHP method is used to determine the weights of the evaluation indexes,and then integrate it with the fuzzy mathematics method to evaluate the health status of the slope.The health level of the slope is obtained according to maximum subordination principle.It is found that the evaluation method can evaluate the health status of the slope well,that is to say,the evaluation model is feasible.Secondly,this thesis uses the dimensionality reduction of the principal component analysis to deal with the factors which affect the health of the slope.It extracts the main factors which affect the health of the slope according to the principal component contribution rate of more than 85%.Finally,it uses the principle elements which are obtained from principal component analysis to predict the slope health status.In this thesis,two prediction models are used to predict the health state of the slope,that's to say,BP neural network prediction model and wavelet neural network prediction model.The principle elements obtained by dimension reduction are bulk density,cohesion,internal friction angle,slope height,slope angle and pore water pressure ratio.In order to reduce the complexity of the prediction model,it uses the principal element as the input of the neural network prediction model.The output of the two models is the safety factor of the slope,which can be used to describe the slope of the health status.The gradient descent algorithm is adopted to modify the network parameters.The only difference is that the wavelet neural network model uses the nonlinear wavelet basis Morlet function instead of the Sigmoid function as the activation function of the hidden layer.The prediction results show that the wavelet neural network overcomes some shortcomings of BP neural network,so that the prediction precision is improved.