Study On Settlement Rule And Prediction Of Metro Station Foundation Pit Surrounding Buildings Based On LSSVM

In order to alleviate the traffic pressure caused by the expansion of urbanization,a large number of subway tunnels and urban overhead are built.However,the foundation pit engineering of subway will affect the safety of the surrounding buildings.Therefore,the deformation monitoring and prediction of these buildings is becoming more and more important.Aiming at this kind of building subsidence prediction problem.Combined with the settlement monitoring data of the building which affected by excavation of foundation pit of Pingdu subway station road in Shanghai metro line 9.Using the wavelet threshold denoising method to deal with sample data,and summarizes the research in the area affected building sink down.Then,the model parameter optimization algorithm GA embedded into the PSO is proposed to construct the new algorithm PSO-GA.By comparing the parameters of the three algorithms,the results show that PSO-GA is more stable and more efficient than PSO and GA.Finally,write the program with Matlab,and the optimal parameters calculated by the three algorithms are brought into the established LSSVM model.Contrast fitting effect and prediction accuracy of building settlement,results show that PSO-GA-LSSVM model fitting effect is better,higher prediction precision,the operation is more stable,reduce into local optimal solution.And prove the improved least squares support vector machine regression prediction model can well predict the deformation trend.Aiming at the research of deformation prediction,it has important academic value,engineering significance and social benefit,which can be widely used.