Study On The Structural Damage Identification Based On The Acceleration Response And BP Neural Network

During the serving period of large and complex structures, the erosion of the natural environment,various disasters and other complex factors will lead to the structure damage with different degrees.Therefore, the structural damage identification, proper evaluation and repair are of great signifince to reducing loss of lives and properties.In recent years, some scholars have proposed to some identification methods based on the modal parameters, but modal parameters are much affected by the environment (such as temperature, humidity, etc.), the incompleteness of the test information and the uncertainty of the structure itself, so they are rarely used in practical engineering; Some scholars have also proposed other structural damage identification methods based on the non-modal parameters,but those methods use the artificial excitation as input.Large-scale structures such as offshore platforms are complex, bulky, expensive,in poor environment,and the artificial excitation are difficult to impose on them,and ambient excitation is difficult to test.So,in this paper, the acceleration response and the BP neural network are used to identify the structural damage of offshore platform.(1)In this paper, two node acceleration response of the structure under the white noise is directly used to calculate the virtual impulse response function, and its amplitude is decomposed by wavelet packet to calculate the node energy.The wavelet packet node energy variation before and after the structure damage is used as the damage characteristic vector, and the pattern recognition function of BP neural network is employed to carry out the structural damage identification,meanwhile, the impact of the different degrees noise on the damage identification is considered. The numerical simulation results of single and multiple damages of the offshore platform show that the method is feasible and has better noise robustness when the structural damage extent is comparatively large. Offshore platform model experiment verifies the validity of the method.(2)In this paper, two node acceleration response of the structure under the white noise is directly used to calculate the virtual frequency response function, and principal component analysis(PCA) is pursued to the amplitude of the virtual frequence response function for dimensionality reduction. The combined principal component variation obtained from the different nodes of before and after the structure damage is used as the damage characteristic vector, and the pattern recognition function of BP neural network is employed to carry out the structural damage identification,meanwhile, the impact of the different degrees noise on the damage identification is considered. The numerical simulation results of single and multiple damages of the offshore platform show that the method can identify the different degrees of structural damage and have some extent noise robustness. Offshore platform model experiment verifies the validity of the method.