Research On Optimizing Combinations Model Of Deformation Prediction

With the development of national economy, people put forward higher requirements for thescale,degree of difficulty, shape of modern engineering structures. In order to master thegeological structure deformation situation of various buildings and natural objects,identifyproblems and take corresponding measures timely,people need to use the limited observationaldata to predict deformed body.Each single model has certain limitations,and the combinationforecasting model has great advantages. But not just any model can be freely combined,excessedprediction model may reduce the prediction accuracy.So in the process of combination,whatprinciples to follow, then make combined model can accurately reflect the actual observationaldata.This paper mainly studies that before building the combination model, first of all,it use thegrey correlation analysis principles and the cointegration theory to analyze predict sequence,andfind out relevancy and cointegration between them, thereby remove the individual predictionmodel which correlation is minor and predict sequence is not smooth.Then according to thedetermined theorem of superior combination, it weed out those redundant prediction methods.Then it simplify steps to create a combination modelThrough case studies and verification of accuracy of the evaluation system,the combinationmodel after optimized selection, its error sum of squares is0.0761mm that less than theminimum error in the single prediction models. It improves the accuracy of the prediction model.whose prediction result is more ideal and reliability is more higher.It can predict the dam,landslide and other deformed body, and can effectively solve the the limitations of individualmodels.