Entwicklung einer numerischen Methode zur Identifikation von Bauwerk-Dämpfer-Systemparametern

• Development of a numerical method for the parameter identification of structure-damper systems

Schleiter, Simon; Klinkel, Sven (Thesis advisor); Kaliske, Michael (Thesis advisor); Zabel, Volkmar (Thesis advisor); Altay, Okyay (Thesis advisor)

Aachen : RWTH Aachen University (2021, 2022)
Book, Dissertation / PhD Thesis

Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2021

Abstract

This thesis is concerned with the parameter identification of structure-damper systems. Hence, it addresses the research areas of structural health monitoring and structural control. The structural health monitoring is part of the system identification, which covers the identification of dynamic properties as well as structural damages based on noisy measurement signals. Due to its recursivity, Kalman filter methods enable a real time identification of systems. Corresponding extensions, such as the Unscented Kalman Filter (UKF), can handle nonlinear systems and in case of the Unscented Minimum Variance Unbiased Estimator (UMVU) also the identification of unknown input signals. Generally, the UKF is only suitable for the identification of continuous nonlinear system changes, such as cracking. However, a main objective of this thesis is to identify sudden system changes, such as sudden failure of structural components induced by earthquakes. Therefore, an adaptation criterion for the UKF is developed that identifies sudden system changes in real time and is statistically independent of measurement noise. Complemented by a localization and a covariance adaptation, it is stated as an adaptive UKF (A-UKF). Numerical simulations for multi-degree-of-freedom systems with vibration dampers are subsequently performed under earthquake, white noise and impulse excitation and show the identification of sudden stiffness changes. Since the related high damping ratios complicate the system identification, the thesis focuses on the identification of such systems. Therefore, parameter studies investigate the influence of the state covariance on the filter sensitivity as well as the influences of system noise and model discretization on the accuracy of the A-UKF. In addition, experimental studies demonstrate the application of the A-UKF to a real multi-degree-of-freedom model with sudden stiffness changes. The second part of the thesis focuses on structural control and puts forth the use of damper systems to reduce structural vibrations. Semi-active dampers, in particular, can respond to changing external conditions by controlling damper properties and thus mostly outperform passive dampers. In order to consider also nonlinear (sudden) system changes a combined method of system identification and semi-active control is developed in this thesis. In particular, the A-UKF is combined with a variable stiffness damper and an associated control scheme is developed, which is referred as variable stiffness control (SVR). It uses a gradient descent method to optimize an energy-based cost function. In addition to the combined method of A-UKF and SVR, a combination of UMVU and SVR is subsequently investigated. Different than the A-UKF, this approach operates without monitoring the input signals and can still control structural response in case of sudden system changes. Numerical simulations under harmonic, white noise and earthquake excitation are performed for both combined methods. Based on the system identification, the damper stiffness is adapted successfully for sudden stiffness changes, frequency de-tuning of the damper and for resonance excitation. A superior efficiency of the semi-active control is observed in comparison to the passive control.