The Research On Damage Identification Of Frame Structure Based On Model Analysis And Neural Network Technology

Due to the influence of own defects, load and external factors, the structural systemwill produce different magnitudes of damage when it is on period. Combined with thereasoning of the motion equations of structure system, changes of the structuralsystem’s modal parameters coincide with the damage occurs. Nowadays, it becomes ahot topic that the structural damage identification based on neural network technologycombined with the structural modal parameters. On the basis of the existing researchresults, this paper mainly contains the following six aspects：1．First, the principle and processes of structural health monitoring technology arereviewed, while the current identification methods of structural damage at home andabroad are summarized. Then, the superiority and inferiority of each method areanalyzed, among which the main point is the development application and researchstatus of the neural network’s applicability on structural damage identification.2．The basic principles of the structural dynamic damage identification methodbased on kinetic theory, which explain the building method and the theoreticalderivation of the damage index base on parameters of frequency, displacement, strain,curvature, flexibility and strain energy, is introduced.3．Based on discusses of the principle and application of the neural network method,the basic ideas and advantages of it for structural damage identification are introduced.Primarily, the technical process of BP and RBF networks are deeply studied.4．Structural damage identification system based on neural network methods isestablished. For reinforced concrete frame structure, the analysis and optimization ofdamage warning index, damage warning network, damage identification index, damageidentification network, anti-noise optimization and measurement point optimization are studied.5．Using the damage identification system proposed in this paper, a specimen offramework instance is tested, and the result shows good identification effect whichmeets the project accuracy requirements.6．Finally, the paper summarizes the problems to be solved in the structural damageidentification using neural network method and the development prospects.This paper lays the foundation of the research on the neural network technologycombined with the structural modal parameters for damage identification, which ishelpful for the future research.