Study On Numerical Simulation Of Diversion Channel And Inversion Of Soil Parameters

China has a vast territory,numerous rivers and lakes,and abundant water resources,but the per capita water resources are seriously inadequate.At the same time,due to the uneven distribution of regional water resources,water diversion project as a comprehensive water conservancy project to effectively alleviate the shortage of water resources is particularly important.Usually,the diversion project is huge and involves many kinds of engineering buildings.How to ensure the safe and stable operation of the diversion project is a difficult problem.Channel is an important part of water diversion project.It is helpful to ensure the safe operation of water diversion project to master the actual state of the structure,slope and foundation of water diversion channel by numerical simulation.However,reasonable and accurate calculation parameters are the important foundation to achieve accurate and effective numerical simulation calculation.Therefore,how to effectively and accurately obtain the relevant numerical calculation parameters of diversion channels is of great significance.On the basis of summing up the previous research results,this paper introduces the basic theory of finite element method and the parameter inversion method of artificial neural network in detail.The finite element calculation model of diversion channel is established by using ANSYS finite element analysis software.Based on the field measured data,the relevant parameters are inverted by using RBF neural network through numerical simulation.In the process of parameter inversion,a single point method is used to invert the elastic modulus and Poisson's ratio of the upper and lower layers of soil.In the construction of learning samples,several groups of parameter schemes are selected according to the actual situation of the project.By changing the combination of different parameters,the settlement of observation points is calculated by using the finite element model of diversion channel.Based on the relationship between water level change and soil settlement,the input and output of learning samples are constructed by using settlement data.Then,according to the monitoring data of the measured points,the parameters of soil are inverted and the inversion effect is analyzed by RBF neural network model.Finally,the structure of the diversion channel is calculated and analyzed according to the inversion parameters.Then,considering the comprehensive application of multi-point information,the inversion of soil parameters is further studied by multi-point method.According to the different sample size of learning samples,it is divided into two ways.The former keeps the overall sample size consistent with that of a single point,and the latter keeps the sample size of each point consistent with that of a single point.Through the above two methods,the RBF neural network is used to carry out parameter inversion of Multiobservation points,and the inversion effect of Multi-observation points is compared and analyzed with that of single observation point.The results show that,compared with single observation point,the parameter inversion of Multi-observation points effectively reflects the relationship between water level change and soil settlement by synthesizing the information of different observation points,and its inversion effect is better.Better results,too.In this paper,RBF neural network is used to inverse the related soil parameters of diversion canal through single and multiple observation points,which can improve the rationality and validity of numerical simulation calculation of diversion canal,and provide guarantee for accurately grasping the safe and stable operation of diversion project.