Structural Damage Identification Of Offshore Platform Under The Influence Of Environmental Factors

The practical application of vibration based structural damage identification methods are influenced by many uncertainties, such as noise interference, data incompletion, environmental factores change, and so on, which reduce the reliability of results of damage identification. In this paper, the research of the structural damage identification under the influence of environmental factors(temperature, mass loading) is investigated. The first-order autoregressive(AR) model coefficient of the measured acceleration response is selected as the damage feature(DF). Based on the fact that local damage will interrupt the interrelation between the DFs from different nodes of the structure, the techniques of cointegration, principal component analysis and factor analysis are employed to eliminate the influence of environmental factors on structural damage identification of the offshore platform. The availability of the proposed methods is verified by the numerical simulations and experimental tests of a five-floor jacket offshore platform model. The main contents are as follows:(1) The vibration based structural damage identification methods are reviewed, and the methods used for eliminating the influence of environmental factors are introduced mainly.(2) The linear cointegration procedure is employed for the DF series from different nodes of the offshore platform, the cointegration residuals which are considered to have eliminated the influence of environmental factors are used as the damage indexes, then, the statistical control chart is employed to identify damage.(3) The principal components and the minor components are used to eliminate the influence of environmental factors on damage identification separately.For the method utilizing principal components, firstly, the original DF vecors are projected into the low dimensional space which is more easily to be influenced by the environmental factors, after that, which are re-mapped back to the original space. The Mahalanobis distances from the re-mapped DF vectors to the original DF vectors are used as the damage indexes which are considered to have eliminated the influence of environmental factors, then, the statistical control chart is employed to identify damage.For the method utilizing minor components, only a few orders of minor components which are insensitive to environmental factors but sensitive to damage are extracted. The Mahalanobis distances from the mimor component score vectors to the center of reference sample are calculated and used as the damage indexes which are considered to have eliminated the influence of environmental factors, then, the statistical control chart is employed to identify damage.The identification results of the aformentioned two methods are identical when the selected orders of principal components and minor components are complementary(The sum of them is equal to the entire orders).(4) The factor analysis technique is employed to extract the common factors of the original DF, and the common factors which are more easily to be influenced by the environmental factors are selected to estimate the original DF vectors. The Mahalanobis distances from the estimate values of DF vectors to the original values of DF vectors are used as the damage indexes which are considered to have eliminated the influence of environmental factors, then, the statistical control chart is employed to identify damage.