Search

In this research, to identify the damage of the nonlinearity system under ambient loads, an intelligent damage identification method based on long-term monitoring data is proposed. The random decrement technique and the autocorrelation function algorithm are used to extract free decay of the structure from long-term monitoring data. The random decrement signatures, autocorrelation function, the frequency of free response and the peak points of the frequency spectrum are used as the features of the structure. These features are then input into the Support Vector Machine (SVM) to classify the current state of the system and their identification accuracy is compared. The simulation experiments results show the extracted features are capable of representing the changes of the system inherent characteristics. Finally, the proposed method is applied to the data analysis of the soft yoke single point mooring (SPM) tower system, and provide the reference for the damage identification of the soft yoke SPM tower system.

Flow-induced motions (FIM) can cause fatigue damage to the mooring and riser systems of deepwater offshore structures, especially for cylindrical platforms, such as Tension Leg Platform(TLP). Previous experimental and numerical studies have showed conservative FIM prediction compared to the field measurements, due to many significant factors such as the Reynolds number, the nonuniformity of the incident current, the unsteadiness of the current velocity and platform heading, the coupled current and wave presence. Besides, the appendages covering the hull surface of platforms also played an important role in the smaller FIM responses of the field measurements. In order to specifically examine the influence of position of pipe-type appendages, this paper experimentally investigated the FIM of a TLP with different positions of appendages at 0°-incidence and compared the FIM responses, energy transformation, and average drag coefficients with those of the bare hull. An air-bearing system and a horizontal mooring system were employed to provide vertical pretension and horizontal restoring forces, respectively, and to ensure the low-friction horizontal motions. The results showed that for 0°-incidence whatever the appendages position, the transverse and yaw motions were all mitigated and the TLP with appendages-45° was the most effective configuration, with a reduction of 55% for the transverse motion and 52% for the yaw motion, mainly owing to the changes of the boundary layer separation point. However, the analysis of energy transformation indicated the development trend of the galloping phenomenon for the TLP with appendages-45°. In addition, the influence of the FIM behavior on the drag force of the TLP at 0° 22.5° and 45°-incidences was also studied. It showed that the FIM could contribute to larger drag force and the increase of drag force is the most significant at 0°-incidence, with a value of around 39%.

During the long term monitoring of the structure, the damping ratio reflects the characteristics of the structure from perspective of energy loss, and its changes can reflect the structural damages to some extent. But in the structural modal analysis based on the prototype measurement, the damping ratio identification results are difficult to identify, especially for the non-stationary structural responses, besides, ICA loses accuracy in the presence of higher-level damping. In allusion to these problems, a modal identification method based on the stationary filter-time frequency independent component analysis (filter-TFICA) is proposed. First a stationary filter using moving average is used to eliminate the non-stationary components. Then the modal identification of mooring system is incorporated into the blind source separation formulation where TFICA is introduced. The validity of the proposed method is confirmed through the identification of a multi degree of freedom numerical simulation system under non-stationary random excitations. Further, the prototype measurement data of the floating production, storage and offloading (FPSO) single point mooring system are analyzed. Compared with the traditional method, the differences of identified damping ratios at the same frequency range by the proposed method are smaller. The distribution of identified modal frequencies is more scattered and the mutually coupled modes are decoupled well. Each of the calculated motion tracks corresponding to the mode shapes presents single form of motion more prominently.

In vibration-based structural health monitoring (SHM), if the data are collected at a higher sampling rate from sensors, a large amount of computing resources and space resources will be consumed. For offshore platforms with limited space resources, there are many difficulties in long-term continuous monitoring. Aiming at this problem, the influence of sampling rate on structural damage identification is studied by comparing and analyzing the acceleration and posture sensors commonly used in offshore platform monitoring. In the simulation of parameter change of nonlinear system, random decrements and their spectrum characteristics are extracted from acceleration and displacement data respectively under different sampling rates, and the characteristics are classified by support vector machine (SVM) to identify the rule of change. The research of model experiment data also shows that the low sampling rate data can realize the identification of the damage, and the recognition degree of the low sampling rate data based on posture is not lower than that of the acceleration. The method is applied to the monitoring data analysis of FPSO soft york single-point mooring system and the damage situation is identical with the actual situation. The results show that the structural damage can be reflected from the low sampling rate posture monitoring data.