Research On Multi-Point Non-Sequential Deformation Forecast Model Based On BP Artificial Neural Network

The analysis research of deformation on deformation body and deformation forecast is an important content of the deformation monitoring, It has important significance for the construction safety and operation on construction projects. However, in the modeling process of practical engineering deformation forecast, often encounter complex deformation forecast problems, such as multi-point, non-linear, non-sequential, etc. To solve this problem, this thesis respectively establish the non-sequential deformation forecast model based on regression analysis model and BP artificial neural network model, and some related problems also have been discussed.The whole thesis is divided into four chapters. Chapter one introduces the aim of deformation forecast, the research status of deformation forecast and some common deformation forecast models, and analyzes the applicability, advantages and disadvantages of common deformation forecast models according to this thesis research content. Chapter two discusses the regression analysis model, the method and the process of establish deformation forecast of the regression analysis model. Chapter three detailed discusses artificial neural network model, the method and the process of establish deformation forecast of the BP artificial neural network modal. Chapter four with the background of project example, firstly Establishing single-point deformation prediction of the regression model, through the analysis of the results and the precision, discusses the necessity of establish multi-point deformation forecast model, then discusses the feasibility of establish multi-point non-sequential deformation forecast of BP artificial neural network model, and analyzes the modeling results and the precision, finally, gets the applicable condition, advantages and disadvantages between deformation forecast of regression analysis model and deformation forecast of BP artificial neural network model.This thesis focuses on the theory and the method research the feasibility of establish multi-point non-sequential deformation forecast of BP artificial neural network model. The multi-point non-sequential deformation forecast of BP artificial neural network model in this thesis could finely solve the problem of multi-point non-sequential deformation forecast, the model have highly forecasting precision and wide application prospect, it has certain directive significance for the further research of multi-point space deformation analysis and prediction problem.