Research On Modeling Method And Precision Analysis Of Deformation Prediction Combined Model

Combined model can preferably fit and predict to the changes of the deformable body due to take full advantages of effective information of each model, also can maintain higher accuracy to reflect the variation of the deformable body effectively. So it always is the research focus of the predict field. In recent years, with the development of science and technology, combined model is broadly used to deformation monitoring domain, provided the effective guarantee for scientific management and security operations of buildings or structures.From the perspective of the existing research results, the establishment of combined model exists in two key issues:(1) the form of combined model;(2) under the condition of any constraints, obtain the optimal solution for the combined model. This paper as a starting point study it in depth on the basis of the deformation characteristics of the data changes, make it as some signals, in general will divide combined model into two modes:(1) tandem combined model based on decomposition of the distortional signal;(2) parallel combined model based on comprehensive utilization of information. This paper studies the two models, which focus on the theoretical modeling and algorithm. The concrete research contents are as follows:This paper summarizes the research at home and abroad and relevant basic theory of combined model, analysis and research three single models:non-equal interval weighted GM (1,1) model、autoregressive AR (p) model and BP neural network model which are related to combined model, and discusses the advantages and disadvantages of the three models. It study the modeling methods of tandem combination, establish the residual error corrective tandem combined model based on non-equal interval weighted GM (1,1) model and BP neural network model. It analysis and research the modeling theory of parallel combinations, systematically study the relevant weighting method, focus on the research in the combined model based on entropy weight method and the optimal weighted geometric average combined model. It analyzes and summarizes problems of the traditional weighted combination empowerment approach, proposes the modeling method of parallel combination based on the Induced Ordered Weighted Geometric Averaging (IOWGA) operators, focus on the research in the definition of IOWGA Operators, the modeling process of combined model based on IOWGA Operators, the determinate method of weight coefficient, the forecasting process of combined model. Taking the subsidence monitoring data of a tailings dam as example, it verifies and analysis the accuracy of modeling methods based on tandem combination and parallel combinations, which reflects the effectiveness of the modeling method.