Back to results list
Please use this identifier to cite or link to this item:
|Title:||Semiactive vibration control using self-sensing magnetorheological (MR) dampers||Authors:||Chen, Zhaohui||Keywords:||Vibration.
Structural control (Engineering)
Hong Kong Polytechnic University -- Dissertations
|Issue Date:||2011||Publisher:||The Hong Kong Polytechnic University||Abstract:||Flexible structures in civil engineering have little damping capability and are sensitive to environmental disturbances. The protection of their structural functionality and safety against hazardous vibration has been gaining significant research and engineering concerns. Particularly, a wide variety of structural control systems has been developed and implemented for that purpose. The research presented in this thesis is devoted to investigation of semiactive vibration control using a self-sensing magnetorheological (MR) damper for a specific kind of flexible structures cables in cable-stayed bridges. To facilitate the real-time closed-loop feedback control in a relatively reliable, simple and cost-effective manner, a smart MR damper capable of force and motion sensing, i.e. self-sensing MR damper, is firstly developed in the form of integration of a prestress-type piezoelectric force sensor and an electro-mechanical transformer with an actuation-only MR damper. The resulting device is in a high degree of sensor-damper collocation and possesses of attractive functionalities of reliable sensing while controllable damping. The self-sensing MR damper is characterized to be a highly nonlinear device with hysteresis and saturation behaviors. A Bayesian black-box modeling technique is developed to formulate its forward and inverse nonlinear dynamics with the synthesis of NARX (nonlinear autoregressive with exogenous inputs) model and neural network, i.e. NARX network, within a Bayesian inference framework. It is indicated that an appropriate architecture design of NARX network is prerequisite and important for obtaining models of simple topology, good accuracy, improved generalization, and fast execution. Furthermore, the Bayesian regularization in an automated fashion incorporated in the network learning is of substantial benefit to reduce the chance of model overfitting and enhance the generalization capability. Accordingly, using experimental data, the Bayesian NARX network models for both forward and inverse dynamics of the self-sensing MR damper are demonstrated to be effective and competent for control formulation, analysis, and application.
To exploit self-sensing MR dampers on vibration control of bridge stays, an adaptive frequency-shaped linear quadratic Gaussian (LQG) based semiactive control strategy is formulated in the frequency domain to have the flexibility of explicitly or implicitly taking into account system uncertainties, unmodeled dynamics and unknown excitations, which guarantees the robustness of the control system. Self-tuning of the regulator gain for adapting to the transient structural dynamics is accomplished by incorporating the Hilbert-Huang transform into the control formulation to capture the instantaneous frequency information of the system. This adaptivity eschews the a priori trial and error on deciding the constant weights usually carried out in the standard LQG case. The forward and inverse models of the self-sensing MR damper are further accommodated in the control formulation to compensate for its nonlinearities. The proposed inverse dynamics based frequency-shaped LQG (iAFLQG) control strategy is numerically demonstrated to be effective in reducing the cable response subject to random excitation, and be superior to the optimal passive control and the standard LQG control with state weight designed to penalize the cable vibration energy. Experimental investigation has been conducted on a 24.2 m inclined cable equipped with a self-sensing MR damper operated in passive and semiactive control modes. Four semiactive control strategies, established based on the standard LQG and the adaptive frequency-shaped LQG (AFLQG) synthesized with a clipped type or inverse dynamics based damper compensator, are adopted in the digital control experiment. The self-sensing MR damper operated in semiactive control mode always outperforms that in optimal passive mode with enhanced cable damping and reduced cable response, since the control strategies are designed to exploit the intrinsic features of the damper in a more efficient way. Moreover, this efficacy evidences the practicability of all the control schemes by using only the cable displacement at the damper installation position and the damper force both monitored by the self-sensing MR damper as feedback information. More important, the superior control performance of the inverse dynamics synthesized AFLQG strategy is experimentally verified. It takes good advantage of the negative stiffness effect yielded by the self-sensing MR damper to remove as much as twice more energy than the optimal linear viscous damper and result in enhanced reduction of cable response beyond the optimal passive control of this damper. In summary, the research addressed in this thesis mainly contributes to the development of a semiactive control system with adaptivity and robustness for protection of flexible structures using collocated smart MR dampers. Aiming to narrow the gap between the time-domain and frequency-domain controller designs, the formulation in frequency domain is more flexible and potential to account for the open concerns in structural control, such as system uncertainties, unknown disturbances, measurement limitations, actuator dynamics, etc.
|Description:||xxxi, 299 leaves : ill. ; 30 cm.
PolyU Library Call No.: [THS] LG51 .H577P CSE 2011 Chen
|URI:||http://hdl.handle.net/10397/4935||Rights:||All rights reserved.|
|Appears in Collections:||Thesis|
Show full item record
Files in This Item:
|b24624883_link.htm||For PolyU Users||162 B||HTML||View/Open|
|b24624883_ir.pdf||For All Users (Non-printable)||8.53 MB||Adobe PDF||View/Open|
Citations as of Oct 15, 2018
Citations as of Oct 15, 2018
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.