Research On Damage Identification Method Of Offshore Platform Structure Based On Long Term Monitoring

In the ocean climate is changeable and the environment is complex,The offshore platform structure is more susceptible to corrosion,fatigue and hidden damage,which reduces the service life of the structure,and even causes structural damage and failure,resulting in loss of production and life.In order to understand the status of the ocean environment and the platform structure In a timely fashion,it is necessary to perform structural health monitoring(SHM)on the offshore platform structure,that is,to monitor the ocean environment and structural prototype on the spot,to find out and warn the problems in time.In view of the structural health monitoring methods of offshore platforms,the following contents are studied in this paper.(1)Characteristic recognition of offshore platform structuresIdentification of structural characteristics of offshore platforms is the basis of damage identification.For non-stationary environmental loads,how to extract structural modal parameters from offshore platform structural response data is studied.A single-sided moving average filtering method is proposed to extract the low-frequency characteristics of offshore platform structures from the filtered random decrement signatures,which has better accuracy.The method is applied to the full-scale monitoring data analysis of FPSO soft yoke single point mooring structure,and the modal frequencies,damping and mode shapes are extracted.Aiming at the snake-like motion of FPSO relative yaw,a method of complex modal parameter identification of ocean structure based on intelligent search technology is proposed.The method constructs an atomic library,which takes on the form of free vibration.Genetic algorithm is used to search the random decrement signatures of structural response data and match them with the corresponding atoms to extract structural characteristics.The correctness of the method is verified by the simulation of multi-DOF system,and the feasibility of the method is demonstrated by the analysis of the monitoring data of soft yoke single point mooring structure.A nonlinear structural analysis method based on acceleration monitoring data is proposed to identify the modal characteristics of jacket anti-icing platform structure under non-stationary sea ice load.The transient characteristics of the structure are extracted by piecewise linear stochastic reduction technique.Several typical time-series data are analyzed to study the jacket anti-icing platform structure And the rule for characteristic changeAiming at the time-varying of soft yoke single-point mooring structure,a method based on EMD decomposition and TVAR model for modal identification of monitoring data is proposed to analyze the characteristics of soft yoke single point mooring structure.(2)intelligent damage detection based on long term monitoring dataThe ocean environmental load is not a wide-band random process,even for linear systems,the inherent characteristics of the offshore platform structure identified only from a small amount of monitoring data are incomplete,It is impossible to judge whether the structure has hidden damage from the features extracted from the data in a certain period of time.For nonlinear systems,a data segmentation method based on structural response strength is proposed,and a random decrement signatures extraction technique under this segmentation condition is presented.Simulations of nonlinear systems verify the effectiveness of the method.Compared with the natural segmentation method,the intelligent classifier established by this method enjoys a higher recognition rate to identify the damage of FPSO soft yoke single point mooring structure.In view of the shortage of space and computing resources of offshore platforms,which is not conducive to long-term monitoring,the acceleration data and attitude data with different sampling frequencies are compared and analyzed.Through nonlinear system simulation and laboratory model experiment,it is verified that the low-frequency acquisition based on attitude sensor is able to identify the hidden structural damage as well.This study also confirms the correctness of using inclination sensor in FPSO soft yoke single point mooring structure monitoring.(3)Intelligent on-line damage identification and early warning based on long term monitoring data.There are many reasons for structural failure of offshore platforms,so it is difficult to predict the location and situation of most hidden structural damage beforehand.It is of practical significance to find out hidden structural damage in time and avoid the significant loss caused by structural failure.In this paper,on-line damage identification method of offshore structures is studied on the basis of intelligent classification,and an on-line damage intelligent identification method based on long-term monitoring data is proposed.This method continuously extracts characteristics from long-term structural response data and establishes a single classifier by machine learning.Normal structural characteristics should be included in the classifier.Incremental learning,training and updating of classifiers are performed for normal structures characteristics with no inclusion.Thus,the classifier based on long-term monitoring data is an on-line damage identification classifier which is formed by learning,training and accumulating experience from the large data of long-term monitoring of offshore platforms.It can identify the hidden damage of offshore platforms online and warn them.