Filtern
Dokumenttyp
Sprache
- Englisch (3)
Referierte Publikation
- ja (3)
Schlagworte
- Damage detection (3) (entfernen)
The implementation of a continuous dynamic monitoring System in Pedro e Ines footbridge at Coimbra, Portugal, operating since June 2007, enabled to detect different environmental/operational effects on the modal properties, based on appropriate processing of monitoring data collected by a set of accelerometers and thermal sensors.
In order to remove or mitigate such environmental/operational effects with the purpose of damage detection, two different Statistical methods have been adopted, One of them consists in the multiple linear regression (MLR) by performing correlation analysis between measured modal properties and environmental/operational variables. Another is based on the identification of the linear subspace within the modal properties by applying principal component regression (PCR) without using measured values of environmental and operational variables.
This paper presents a comparison ofthe performance of these two alternative approaches on the basis of continuous monitoring data acquired front the instrumented Pedro e Ines footbridge.
In the context of national innovative project IMO-WIND, an integrated long term monitoring System was installed on a prototype of an offshore wind turbine System of 5 megawatt dass. The Federal Institute for Materials Research and Testing (BAM) was responsible for development of the SHM System. It consists of a signal acquisition System and a Signal processing and management System.
The signal acquisition System is composed of 14 acceleration sensors, 110 strain gauges and 4 inclination sensors, installed at the tripod foundation and the tower. It began to work in August 2007. In order to investigate the structural dynamic properties, extract efficient damage index and manage huge amounts of Vibration Signals and analysis results, an automated signal processing and management Software System is developed in LabVIEW® environment. It includes functions such as automated Operational Modal Analysis (OMA) on the basis of Stochastic Subspace Identification (SSI) method and poly-reference Least-Squares Complex Frequency Domain (p-LSCF) approach, investigation of environmental/operational effects on structural dynamic properties under operational conditions, features extraction using Principal Component Regression (PCR) and Multiple Linear Regression (MLR) as well as data/result management and visualization.
Besides, this paper also presents variations of structural modal parameters of a wind turbine system under complex environmental/operational conditions. Comparison of modal parameters automatically extracted by SSI method and poly-reference p-LSCF approach indicates that the latter method provides more stable modal identification results ffom the viewpoint of long term OMA. Subsequently, the influences of temperature, wind velocity, rotation speed and nacelle direction on modal parameters are generally discussed. Finally, variations of indentified modal parameters during two years are presented.
The second part of these companion papers mainly researches environmental/operational influences on structural dynamic properties under normal operational conditions during two years, in order to extract a statistical based damage-sensitive indicator for health monitoring of a wind turbine system.
The correlation analyses between experimental identified frequencies, damping values as well as mode shapes and environmental/operational factors such as rotation speed of blades, wind speed, pitch angle, temperature and nacelle direction are presented. It is observed that the frequency estimates are influenced by the nacelle position, the activation of rotor, the rotation speed of blades and the wind speed as well as the temperature. Regarding to the damping estimates, they are mainly associated with variation of the aerodynamic damping due to the increasing wind speed. Besides, the resonance phenomenon is also observed in higher modes. The harmonic frequencies due to blades passing by tower are found and the corresponding damping value decreases. Moreover, the mode shapes in some modes are strongly affected by the position of the nacelle.
Subsequently, two types of simulated damage including the reduction of stiffness in both the rotor blade and the tubular tower are successfully detected by applying the Principal Component Analysis (PCA) based methods to these temperature-sensitive frequency estimates. Comparison of change of the extracted health features indicates that they are more sensitive with the tower damage.