Research On Damage Identification Of Grid Structure

Due to the rapid development of urban construction,various long-span structures have been constructed and put into use.Among them,the role of grid structure has been gradually paid attention to.The grid structure is composed of rods and bolt balls.These components make the grid structure into an indeterminate spatial three-dimensional structure.However,under the influence of external load,environment and other factors,the members of the grid structure are often prone to fatigue accumulation,material aging and other problems.When the damage accumulates to a certain extent,it will lead to structural damage and cause casualties and economic losses.Therefore,it is necessary to carry out preventive damage identification on the grid structure in service.For this reason,the damage identification method of grid structure is studied and discussed in this paper based on the dynamic test experiment of the full scale grid structure model.The main research contents and achievements of this paper are as follows:A full scale grid structure model of 14m×1m×0.4m was established.Through dynamic testing of the grid structure,data collection of 15 damage conditions was completed,and a damage sample database was established for the damage identification method studied in this paper.The sensor arrangement,excitation position selection and acquisition equipment selection are studied,and a set of practical and feasible experimental steps and operations for dynamic test of grid structure are established.A damage identification method based on modal curvature difference and generalized regression neural network is proposed.Firstly,the mesh damage location is located by the first-order modal curvature difference.Second after normalization of input data,through optimizing the spread value of generalized regression neural network model for identification of grid structure experiment research,the results show that the method can accurately to the single pole of grid structure model,more than two bar and rod is used to identify the position location and degree of damage condition,and the identification accuracy is higher,the recognition accuracy is above 90%,overcomes the traditional damage identification method for long-span grid structure problems such as insufficient recognition accuracy.A method for damage identification of grid structure based on AR feature parameters and support vector machine(SVM)was established.This method firstly constructed a 12-dimensional AR feature parameter graph to represent the characteristics of each group of damage conditions,and then normalized the input data to identify and predict the damage by a SVM with radial basis function as the kernel.Through the study of the grid structure model in this paper,the results show that the method can effectively identify the different positions and degrees of the damage condition of the single,two and three rods of the grid structure,and the recognition accuracy is above 95%.The recognition accuracy is better than the generalized regression neural network,and the recognition effect is good.Moreover,in terms of anti-noise ability,SNR value of support vector machine is lower than SNR value of generalized regression neural network under the condition of ensuring the same recognition accuracy for the same working condition,and support vector machine can guarantee the recognition accuracy in higher noise environment.This method has high practical value in the damage identification research of actual grid engineering.In this paper,based on the research of grid structure dynamic measurement experiment,based on the modal curvature difference is established and generalized regression neural network grid structure damage identification method,the characteristic parameters based on AR and grid structure damage identification method of support vector machine(SVM),two kinds of structural damage identification methods respectively in the grid structure damage condition was validated,damage identification for the actual grid engineering provide certain technical foundation.