Model Updating Substructure Hybrid Simulation Test Method Based On Intelligent Algorithm

In the process of hybrid simulation test,the unknown parameters in the test substructure are identified and updated to the computational substructure to reduce the scale of the test substructure and improve the test precision.Considering the complexity of the node parameters and the important influence on the structure,this paper takes the parameter of the test substructure node as the identification object,using the Elman,GRNN,ELM and SVM intelligent Algorithms,we design the study library,identify the node parameters,update the computational substructure and motion equations and complete the whole hybrid simulation test process through the idea of uniform design.In order to achieve a better recognition effect in the hybrid simulation test,this paper uses the following methods: 1)using Matlab and Open Sees for hybrid programming calculations;2)replacing the original computational substructure model with the overall model;3)improving the uniform design of the table design learning library;4)identifying the parameters in the process of identify the value of the method and the criteria.In addition,using the recognition effect factor to determine the good or bad fit of the model and the real model in the structural response,and then giving the calculation method of the recognition effect factor.This paper establish a model updating substructure hybrid simulation test system based on intelligent algorithm and carry out a lot of numerical simulations on the model.The three-parameter,six-parameter and nine-parameter identification of a steel frame model with a span of five layers are realized under the acceleration of different seismic peaks.From the analysis of the recognition result,we know that the method can effectively identify the structure parameter value;the model also can be fitted with the real model in the structure displacement and force response.In addition,the recognition effect of the parameter value,the established model of the parameter value and the real model have no obvious advantages and disadvantages in the degree of the structural response.This paper uses GRNN intelligent algorithm to further verify the applicability of the method.By setting the initial values of different levels in the range of parameters,it is verified that the initial values of different levels can achieve better recognition effect,and the recognition effect is not affected by the initial value.It is proved that this method can be used to identify the parameters of steel frame model effectively by establishing the model of different layers and different spans.It is also proved that this method can effectively resist the interference of measurement error and the parameter identification by simulating whether the test substructure is affected by the load measurement error of the test equipment during the real test.