Study On Slope Deformation Prediction Of Support Vector Machine Parameters Based On Improved Ant Colony Algorithm

With the rapid development of urbanization process,lots of side slopes emerge in the cities.In recent years,slope instability accidents occur frequently,which undoubtedly brings considerable disaster to the country and the people.Therefore,we need to establish an effective prediction model to predict the deformation tendency of the slope in the future,and finally achieve the purpose of slope deformation research.Due to the traditional way has its limitations,optimize the support vector machine algorithm by combining several algorithms of support vector machine,and combined with the grid method to optimize the support vector machine parameters,finally applied to slope deformation prediction.Firstly,the significance of slope deformation prediction is expounded,and the present situation of slope deformation prediction is analyzed.The support vector machine model which can solve the problem of small sample,nonlinear and high dimension is put forward In slope deformation prediction.Aiming at the problem that the traditional support vector machine is difficult to determine the parameters of the model,the basic ant colony optimization is easy to fall into the local optimum,the two factors and the volatility coefficients in the ant colony transition probability formula are proposed to form an improved ant colony optimization.And the grid method is used to search for the best parameters of support vector machine,Finally,the improved ant colony optimization support vector machine parameter model is established.Secondly,we choose two examples of slope deformation data,take a one-step prediction method,combine the Microsoft Visual C ++ 6.0 compiler on Matlab platform,use the libsvm toolbox for extended programming,to improve the ant colony optimization support vector machine model Training and forecasting.Finally,according to the Matlab program,three kinds of models are used to test the slope data of two slopes in china.The experimental results are evaluated by mean relative error,The average relative error from the genetic algorithm support vector machine model were 6.74% and 6.71%,the average relative error of particle swarm optimization support vector machine model are 4.99% and 4.16%,and the average relative error of the improved ant colony algorithm support vector machine model were 2.60% and 2.28%,and the mean square error and square sum error of the improved algorithm are less than those of the other two methods.The results show thatthe improved ant colony algorithm can optimize the support vector machine model to predict the slope deformation better than the other two models in the paper.