Forecasting Of Slope Displacement Based On MAPSO-SVR With Mixed Kernel

With the emergence of many slopes,landslide events are frequent,resulting in great impact and loss on people's production and life.Therefore,prediction and early warning of slope deformation is particularly urgent and important.In view of the shortage of traditional prediction methods,it is necessary to discuss new methods to predict.In this paper,the improved support vector machine(SVM)is applied to the slope displacement prediction.First,the traditional PSO algorithm is easy to fall into the extreme value,in view of the difficulty of obtaining the parameters of the support vector machine model.The inertia weight in the velocity updating formula of particle swarm optimization with adaptive inertia weight replacement,Using multi particle information sharing to improve the position updating formula of particle swarm optimization,Construction of MAPSO algorithm.In order to solve the problem of generalization and learning ability of single kernel function,the kernel function of Poly and RBF plus descending weight is used instead of SVM kernel function.The MAPSO algorithm is used to optimize the parameter of the support vector machine of the mixed kernel function.Finally,the slope displacement prediction model of the mixed kernel function MAPSO-SVR is established.Secondly,taking Danba mirror landslide displacement data and new Wolong landslide displacement data as experimental data,we use the Libsvm enhancement toolbox developed by Faruto and others,to train and predict the slope displacement prediction model of mixed kernel function MAPSO-SVR.Finally,according to the compiled Matlab program,the prediction results of slope displacement based on the mixed kernel function MAPSO-SVR are compared with the prediction results based on traditional support vector machines.Among them,the Danba mirror landslide displacement data,the maximum and minimum relative error of hybrid kernel slope displacement function prediction of MAPSO-SVR were 6.97% and 0.55%;based on the mixed kernel function MAPSO-SVR and the mean square error and the square error were 0.014 and 0.074;the maximum and minimum relative error of PSO parameters optimization algorithm support vector machine the displacement of slope prediction is respectively 9.92%,2.15%;based on PSO parameter optimization algorithm support vector machine prediction of slope displacement mean square error and square error and were 0.033,0.166.The new Wolong Temple Landslide data,The maximum and minimum relative errors of the slope displacement prediction based on the mixed kernelfunction MAPSO-SVR are 3.67% and 0.16% respectively,based on the mixed kernel function MAPSO-SVR the mean square error and the square sum error of the slope displacement predictionare 1.16,6.90.based on the PSO parameter optimization algorithm support vector machine for slope displacement the maximum and minimum relative error of the prediction is respectively 10.46%,3.79%;based on PSO parameter optimization algorithm support vector machine prediction of slope displacement of mean square error and square error and were 7.35,44.08.It can be verified that the slope displacement prediction based on the mixed kernel function MAPSO-SVR has higher accuracy.