The Combination Forecasting Model And Its Application In Deformation Analysis

Due to the different kinds of factors, the buildings would produced deformation in the construction, operation or management. However, the deformation of the complexity and diversity and this tendency is unlikely to be reflected by the single prediction models. In other words, these single models can not avoid the error, and we all know different single models sometimes could provide something useful information, if we removed the low prediction accuracy forecasting models, there will be affecting the model rationality. Since the combination forecasting model was first proposed in 1969, it can combine the effective information of each single forecasting model and reflect the essence of the things, which has attracted the attention and research of the vast number of Surveying and mapping scholars.In this paper, it is described the connotation of deformation and the commonly used deformation analysis, and the concept of combination forecasting model. According to the characteristics of the deformation forecasting model, these combination model were classified into two types: Tandem and parallel combination forecasting model. This paper research and analysis the combination forecasting model, and details are as follows:1) According to the BP neural network can easy fall into the local extremums, an improved AWSPSO algorithm is proposed and used the AWSPSO to optimize the weights and the thresholds of BP neural network. According to the characteristics of Tandem combination model, a new revision residual error combined forecasting model based on GM(1,1) and improved AWSPSO-BP model.2) Describes the concept and classification of the parallel combination forecasting model, and also systematic study the parallel combination model with certain weights of criteria. This paper introduces operations research of the Grey Correlation and proposed a kind of the Grey Correlation Degree maximum criterion for fixed weight combination forecasting model. In order to overcome the consistency problem of the results of a single evaluation method, it proposed a new combination evaluation method of the minimum loss of opportunity. According to the problems of the traditional weighting method, it proposed a new forecasting model based on the Second-order Forecast Effective Measure and IOWGA Operator.3)Finally, the combined forecasting model is verified by an example. Form the result, we can find this two combination models has the best accuracy.