Study On Damage Identification Of Structures Based On Wavelet Packet Energy And Neural Network

With the development of the transportation enterprise, lots of bridges have been built,while what is worth noticing is the increasing of the number of damage bridges and extend of bridge damage. For some important bridges, along with they work more time, their strength and stiffness must decrease. So the damage detection of bridge structure becomes much more important because of all the problems that whether the load-carrying capacity of old bridges can satisfy the needs and whether the quality of the new bridges can be up to scratch.In the damage detection of bridge structures,the methods of damage identification based on dynamic detection are maturing. However, most methods of measurement based on the dynamic detection need force test,and testing force brings about some problems.This paper is committed to seeking a no force needed test method based on dynamic detection which is useful for reinforced concrete structure . This paper studies the application of wavelet packet energy as well as the wavelet packet energy in the use of neural networks ,in order to identify the location of injury and the extent of injury.Through finite element analysis to prove that the wavelet packet energy can accurately determine the location and extent of damage of bridges. In the laboratory , we establish a 5.3m reinforced concrete simply supported T beams, through the classification of different injuries have been loaded. Through the impact of the constant using of homemade device, we pick up different damage in a state of acceleration signal. Through the wavelet packet energy rate index,we can accurately determine the damage location of the reinforced concrete structure.We use the wavelet packet energy distribution vector as input,the degree of damage as output, establish RBF network can accurately determine the extent of damage. The two-step identification method raalizes of the first to determine the location of damage, after damage to determine the extent, greatly reducing the required sample of neural networks.In this paper, confirmed the effectiveness that we can identificate the damage of reinforced concrete structures with unexpected power wavelet packet energy , provied the theory and experiment of new reinforced concrete structure damage identification methods.