Research On Several Improved Deformation Prediction Model

Deformation is the normal state of existence in nature,it is closely related to our daily life.Deformation has an imperceptible impact on the safety of people's lives and property,as well as the national infrastructure,each year,the loss due to deformation is immeasurable all over the world.Therefore,understanding the deformation,analyzing the deformation,grasping the deformation law,and predicting the deformation trend are of great significance in reducing the occurrence of major accidents and reducing the loss of manpower and material resources.However,the causes of deformation are complex and diverse,there is no model that can accurately describe and predict the deformation law of deformation bodies.In view of the above situation,many scholars at home and abroad have done a lot of research work on deformation prediction models and their combination models,they look for the causes of deformation of deformation bodies through different approaches and methods,and then establish corresponding models to simulate the law of deformation occurrence.,and achieved a good results.Although there has been some progress in the research and exploration of deformation forecasting models,the phenomenon that the accuracy of deformation forecasting models is generally low still exists.How to improve the accuracy of deformation forecasting model becomes a hot and difficult issue.This paper focuses on the key modeling problems in deformation analysis and forecasting and introduces advanced mathematical theory and analysis methods to improve the deformation forecast model from multiple angles,and through engineering examples,to analyze and compare the prediction accuracy and applicability of the improved model.1.For the actual situation that there may be small samples?non-equal spacing observation or Some period of missing,and the large deformation monitoring points of the buildings are mutual influence and mutual correlation.a method to extend the non-equidistant single-point deformation prediction model into non-equal spacing multi-point deformation prediction model is proposed,and then according to the characteristics of the non-equal spacing multi-point deformation prediction model,from three angles that correct the prediction results of the model,improve the original modeling data sequence smoothness and optimize the model parameters,the Markov theory,the wavelet denoising principle,and the multiple integral whole least squares method are used to improve the model.Combined with examples to verify the accuracy of the improved model.2.The modeling method and theory of the nonlinear grey-time series GM-AR combined forecasting model are studied.The method of introducing the global least-squares(TLS)method to optimize the parameters of the combined model to improve the model forecasting accuracy is discussed.3.For the characteristics of high-speed iron settlement,the optimization forecasting algorithm is considered from multiple angles and combines the characteristics of wavelet analysis theory,least squares support vector machine prediction model,and AR time series analysis model,an LSSVM-AR combined prediction model based on wavelet analysis was established.Combined with the settlement data of high-speed railway foundations to analyze and compare,the prediction accuracy of the combined model is verified.