Research Of High-speed Railway Subgrade Settlement Prediction System Based On Intelligent Computing

The settlement of subgrade is the key issues involved in ballast less track ofhigh-speed railway in the process of building, The sub line of the settlementprediction of roadbed engineering is a key technology in the construction processHigh-speed Rail roadbed, It is directly related to the safe operation and servicelife of high speed railway. In this paper the existing subgrade settlementprediction model is proposed based on the combination forecast model, and usedfor high-speed railway subgrade settlement prediction, it has good predictioneffect.The paper analyses the existing single forecasting model for subgradesettlement of high speed railway, such as Three point method, Asaoka method,hyperbola, GM gray prediction, Verhulst gray prediction method, and throughsimulation experiments on subgrade settlement prediction effect comparison. Theestablishment of the subgrade settlement prediction model using least squaresupport vector machine, the sample data respectively by "time-sedimentationvalue" series and "settlement value-sedimentation value" sequence manner,improve the prediction accuracy of forecasting models, at the same time usingrolling window form, increase the stability of the forecasting model, and thepredicted effect of the simulation and verification.In order to simplify the combination forecast model structure, improve theprediction accuracy, the paper puts forward the application of cointegrationtheory to various single forecasting model selection, and combined with theredundant information screening principle, selection to meet the cointegrationrelationship, and contains a useful predictive information forecasting model.The establishment of combination forecasting model. The forecasting modelare selected respectively establish fixed weight combination forecasting model and variable weight combination forecasting model, fixed weight combinationforecasting model is based on error square and minimum principle, several kindof single phase selected prediction modelaccording to a certain weight,combination output after subsidence prediction model for output value fixedweight combination; variable weight combination forecasting model is top redictthe single forecasting model selected values as inputs of support vector machineoutput support vector machine as the forecast output model. The combinationforecasting model is effective and integrated all kinds of forecasting methods ofinformation, improve the prediction accuracy at the same time, also caneffectively reduce the risk prediction.Using the method of verification of Lan-xin high-speed railway and Zheng-xi high-speed railway subgrade settlement of actual data, a simulation exampleshows that the prediction effect model not only improves the prediction accuracycomposite subgrade settlement of high speed railway proposed, but also reducesthe prediction risk, improve the anti-interference ability, has high practicalapplication for engineering reference value.