| With the processing and installation techniques of grid structure increasing maturity, equipment upstairs has been more common, especially in areas such as the clinker library in cement industry, prehomogenization libraries and aerospace A380 hangar, weight of the equipment become more heavy and load become more complex. Long-term operation of these devices can cause structure damage such as partial bolts looseness and rods bent and broken, which can easily lead to catastrophic incidents. Damage detection of grid structure and taking preventive measures are particularly important. However, due to there are at least a thousand nodes many as ten thousand for the grid structure, it’s so difficult to find the damage by artificial observation or detection. And the grid structure is used in the roof part of the structure; the height difference also makes it difficult to use the instrument to detect damage. Therefore, research on the damage identification of grid structure in service has important practical significance.In this paper the method based on wavelet packet energy rate and the method based on FRF and principal component analysis are use through numerical simulation. The two methods are all based on the time-domain signal of the structure measured by hammering method. The first method is taken out through wavelet packet toolbox in Matlab, by which the packet energy value of each band for each node of the grid structure. Then compare the four conditions of undamaged, single place injury, two injuries and two injuries having no common node. Select the wavelet packet energy of before and after damage as the structural damage index to detect the damage location. By the second method the FRF can be obtain from the time-domain signal. Because the dimension of FRF is relatively large, the calculation is more complicated, and therefore combining the principal component analysis to reduce the dimension of the FRF to get the FRF vector. Select the FRF vector angle of before and after damage as the structural damage index to detect the damage location. In this part of numerical simulation all nodes of the grid structure are analyzed, which is not easy to achieve in practical engineering. First, the number of nodes in the grid structure and the number of nodes is relatively large, detecting one by one is relatively a large work. Second, the number of the sensors for damage detection in general is limited. So, it’s difficult to detect the all the nodes of the structure. Therefore, the method of optimal sensor placement based on strain energy participate coefficient is put out in the later section of the experimental analysis, which can reduce the number of the nodes greatly. The number of detected nodes for the first time is relatively small, thus the tow-step method of damage detection of grid structure is required. Two methods are used in the part of experimental analysis, and because of the limitations of the grid structure, only two conditions are used in the experiment.The numerical simulation and the experimental analysis show that the recognition result of the damage detection method based on wavelet packet energy rate and the damage detection method based on FRF and principal component analysis are ideal for single injure and the damage location can be detected accurately. In the numerical simulation for the two damage the corresponding nodes of the damage node can be detected, but the different node with same node number called “suspected node” will appear, which mainly due to the grid structure is generally higher statically indeterminate structure, and the problem should be solved in the future analysis. The two methods are accurate and effective eliminate interference of the “suspected node’. The sensor placement method based on strain energy participate coefficient is necessary to reduce the number of the nodes, and the work efficiency is greatly improved. The detection method can be used in the practical engineering. |
PDF Full Text Request