Optimization Of LSSVM Model Based On Improved Gray Wolf Algorithm And Its Application In Deformation Monitoring

With the rapid development of social economy,urban buildings rise on the ground.Its height is also higher and higher,followed by the deepening of the foundation pit.Deep foundation pit is widely used in modern buildings,and it is more and more needed by people.Due to the movement of the earth's crust and the use of groundwater,subsidence has been taking place all the time.But the settlement of deep foundation pit is related to the foundation of high-rise building.In today's people-oriented social life,safety has always been a top priority topic.The settlement of deep foundation pit is closely related to people's health.Therefore,the observation and prediction of foundation pit settlement has always been the focus of attention.In the process of foundation pit construction,it will certainly cause its own and surrounding turbulence,which will cause deformation of underground tunnels,pipelines and surrounding buildings.Certain deformation can not be prevented by manpower,but exceeding a certain limit will cause great loss to people's property and life safety.Therefore,in the process of foundation pit construction,it is necessary to monitor the deformation of foundation pit,and analyze the causes and laws of deformation according to the deformation data.Try to minimize accidents.In this paper,the machine learning algorithm is used to optimize the least squares support vector machine through the improved gray wolf algorithm,and it is applied to the deformation monitoring of foundation pit to predict the settlement of foundation pit.The main work of this paper is as follows:(1)The deformation factors and conditions of foundation pit during construction are simply understood.At the same time,all kinds of prediction models are classified and evaluated.The least square vector machine is selected as the model of foundation pit prediction.Because it is difficult to determine the parameters of LSVM,this paper proposes a gray wolf algorithm which is easy to understand,has good convergence and strong ability of optimization.(2)For Gray Wolf algorithm,this paper first introduces the origin and principle of Gray Wolf algorithm.At the same time,the disadvantages of Gray Wolf algorithm are analyzed:(1)the initial population is stochastic.(2)There is no better display of the different rank status ofwolves.It is mainly improved by variable proportion weight and adding differential evolution algorithm.After the improvement,four algorithms and six functions are used for performance comparison and analysis.The results show that the improved algorithm not only improves the optimization ability,but also improves the iteration speed.(3)The improved gray wolf algorithm(IGWO)is used to optimize the parameters of LSSVM,mainly the kernel parameters and regularization parameters of LSSVM,and then a new model,IGWO-LSSVM prediction model,is obtained.Then three commonly used detection model accuracy indexes are used to detect the IGWO-LSSVM prediction model.The results show that the IGWO-LSSVM model improves the prediction accuracy and efficiency.(4)With the help of MATLAB language,combined with the actual deformation monitoring data,three models of LSSVM,GWO-LSSVM and improved IGWO-LSSVM are simulated and trained.The accumulated settlement value is predicted and compared with the data obtained from deformation monitoring and observation.The results show that the prediction accuracy of GWO-LSSVM and the improved GWO-LSSVM model is higher than that of LSSVM,and the prediction results of the improved GWO-LSSVM model are closer to the actual observation values.The network model has a good application value in the deformation prediction of urban foundation pit.