Multiple Parameter Combination For Damage Identification Based On The Improved BP Neural Network

In recent years, due to structural damage caused by the frequent occurrence of major accidents, the reasons are mainly structural cracks, caused by aging and damage have not been effective in preventing and controlling. Structural safety is essential, structural health monitoring has become a hot topic in recent years. Structural damage identification as the important part of health monitoring has caused wide public concern. particularly in the damage identification method and accuracy has done a lot of research. In view of the traditional damage identification methods, this paper mainly discusses the application of neural network to identify structural damage location and degree were studied.In this paper, through theoretical analysis it is concluded that the parameter combination is natural frequency changes and some more modal combination as a neural network input can damage identification of structures, And application of the improved BP neural network, to establish the beam by the method of step by step to identify numerical example is used to numerical simulation of the damage. At the same time, to cuntan Yangtze river bridge project for the implementation of the background, puts forward the layered before the damage identification method of step by step, further verified the multi-parameter combination as the feasibility of the neural network input for damage identification.The main contents include:①Analyses the working principle of neural network, this paper studies the improved BP neural network for structural damage identification applications, in theory, it is concluded that the multi-parameter combination as the input of the BP network can be used to identify the damage of structure.②Numerical simulation by finite element software ANSYS to establish simply supported girder bridge, According to the step by step identification method to judge the damage location and damage degree, using three species different of network parameter entered(inherent frequency changes, and some point modal, and Multi-parameter Combination) established neural network, comparative analysis of its effect, Multi-parameter combination to identify the damage can be and overcome the alone natural frequency caused by defects.③Based on the cuntan Yangtze river bridge project for the implementation of the background, considering the complexity of bridge structures,establish multiple neural network using the step by step after the first layer is used to identify the damage identification methods, focusing on main cables of suspension bridge with multiple parameter combination as a neural network is used to identify the location and degree of the input. This method is proved by further engineering practical value.