| In order to ensure the safe operations of the jacket platform,it is important to monitor the structural health status real time.Hence,the structural health monitoring technology has drawn much attention from scholars and the industry.The basic method of this technology is to extract the damage-sensitive features from the dynamic responses,which can reflect the structural changes.Based on these features,a damage index can be developed to estimate the damages.However,many studies have shown that the damage-sensitive features are not only influenced by structural damages,but also by environmental factors.The offshore jacket platform is primarily subjected to random wave excitations,and this type of excitation can be characterized by short-term random fluctuation and long-term random evolvement both.Thus,it is difficult to identify whether the changes of damage-sensitive feature are attributed by the structural damages or environmental variations.Therefore,a damage-sensitive feature is firstly proposed under the framework of model-based method,which is based on the multi-point crossover tensor mapping of the structural dynamic responses.On this basis,the uncertainty analysis method of the data-driven method is introduced to establish a new principal component analysis method.This method can achieve the refining of the mono-damage-sensitive feature.Further,the final damage detection index is proposed by constructing the method of distance measurement between the data samples and the datum reference set.Through the verification for the numerical model of the target platform,this index has the ability to resist the short-term and long-term wave interference both.An important discovery has been made in exploring the background field difference between the virtual and actual monitoring data.The proposed indicator of the noise contamination level can effectively measure the difference between the two background fields.By introducing this difference into the datum reference set for the calibration,the migration of damage detection method is realized from the virtual monitoring data to the actual monitoring data.In order to test the accuracy of the damage detection method,the model tests were carried out.Through the detailed analysis of the test results,it can be found that the accuracy of the damage detection method is depended on the priority of the damaged members.For the group members with the highest and second-highest priority,the accuracy is more than 95%.Through the numerical simulations and model tests,the effectiveness of the damage detection method has been well verified.However,there will be another obstacle when the method is applied to the target platform,which is the short-term interference effect caused by the operational effects on the platform.These effects are often manifested as the instantaneous impulse with large amplitude.On this basis,after fully analyzing the time-frequency fluctuation characteristics of the impulse noise,a wavelet denoising method is established for the field signal based on the kurtosis detection and Parzen window probability density estimation.The effectiveness of this method is verified by the virtual and actual monitoring data.Hence,it paves the way for the practical application of the damage detection method oriented to the target platform.Under the guidance of the proposed damage detection and denoising methods,the field monitoring data from January 2019 is analyzed.Considering the edge effect in practical application,a homogenization smoothing method is proposed.This method can not only eliminate the edge effect,but also maintain the high real-time and dynamic performance of the damage detection.And the results show good stability of the structural state detection.Thus,it can be seen that the damage detection method proposed in this paper has good application effects,which provides a valuable research idea for the damage detection of the offshore platform under the random ocean environments. |
