H. Allison Smith

Vibration control of cable-stayed bridges—part 1: modeling issues

The objective of the research presented here is to increase the understanding of how the complexities associated with modeling cable-stayed bridges, such as non-linear behaviour and the participation of highly coupled, high-order vibration modes in the overall dynamic response, affect the overall effectiveness of active control schemes. The 316-degree-of-freedom analytical model studied here is based on the Jindo Bridge located in South Korea. Computational considerations associated with control analyses require the size of the model to be significantly reduced, without loss of the important vibration characteristics and complexities. Three separate reduced-order modelling techniques for creating effective control models are studied here: the IRS method, the internal balancing method, and a modal reduction method. These methods are studied and compared on their ability to capture the complex dynamic response of cable-stayed bridges subjected to multiple-support excitation and their ability to create viable and computationally sound state-space models for control analyses. Results show that the modal reduction technique, because of the ability to select only those modes causing the largest force and displacement response, is most effective for control applications.

EFFECTIVE OPTIMAL STRUCTURAL CONTROL OF SOIL–STRUCTURE INTERACTION SYSTEMS

A methodology is developed in this paper to include soil–structure interaction effects in optimal structural control, General Multi-Degree-Of-Freedom (MDOF) structural models are considered. The SSI transfer functions for ground motion and control force in the physical space are presented first, followed by a methodology for using system identification techniques to find an equivalent fixed-base model of an MDOF SSI system. An iterative technique is applied to combine these methods for the determination of optimal control gains. The control effectiveness of considering soil–structure interaction is investigated for the controlled SSI system. It is found that the control algorithm considering SSI effects is more effective than the corresponding control algorithm assuming a fixed-base system model. In addition, the advantage of applying this methodology is observed to be more prominent in the cases where the SSI effects are more significant.

Vibration control of cable-stayed bridges—part 2: control analyses

The objective of this research is to increase the understanding on how the complexities associated with modelling cable-stayed bridges, such as non-linear behaviour and the participation of coupled, high-order vibration modes in the bridges dynamic response, affect the overall effectiveness of active control schemes. Using a reduced-order state-space model, control analyses examine the effectiveness of full state feedback control employing a Linear Quadratic Regulator (LQR) and the effectiveness of dynamic output feedback control utilizing a Kalman–Bucy filter in attenuating the structures force time-history response. Results show that significant reductions of the maximum internal forces and the force/displacement response can be achieved through full state or output feedback control. An investigation of various actuator configurations leads to the conclusion that actuators are most effective when located close to the centre of the bridge span. The study also shows that only first-order modes need to be controlled to reduce the displacement response; however, the control of higher-order modes is essential to reduce the force response. Multiple-support excitation needs to be considered since it can excite entirely different modes than uniform-support excitation. Moreover, multiple-support excitation induces forces that are caused by pseudo-static displacements and can not be controlled. Special attention needs to be given to coupled modes since their control can lead to an increased force response of the structure.

Identification of Structural System Parameters Using the Cascade-Correlation Neural Network

he use of neural networks for structural system identification is receiving an increasing amount of attention through the research focused on structural control and intelligent systems. These systems require continuous monitoring and controlling of structural response; thus, on-line identification techniques are needed to provide real-time information about structural parameters. The Cascade-Correlation (Cascor) neural network is applied here to the structural system identification problem. The Cascor network utilizes a dynamic network architecture and a variable error threshold mechanism which facilitates training and can increase the networks ability to generalize

Design of H∞ output feedback controllers for the AMD benchmark problem

This paper outlines a general approach for the design of H∞ dynamic output feedback controllers and applies this method to designing controllers for the Active Mass Driver (AMD) benchmark problem. The example controllers designed for this problem use acceleration output feedback of the structure coupled with the additional actuator sensors and ground motion sensor. Some of the key choices made by a control designer using this method are discussed and evaluated with example controllers. Several sets of controllers are developed to evaluate the sensitivity of controller effectiveness to the choice of regulator response quantities, the choice of feedback quantities, and the choice of when to apply model reduction. Results show that for this design approach the best controller effectiveness is achieved by choosing to regulate the structural accelerations and displacements with the controller acceleration and command signal. In addition, the sensitivity of the dynamic controllers to the removal of available sensors is investigated, showing that the performance of the dynamic controllers for the nominal AMD model are insensitive to which sensors are available.

A Time‐Domain Algorithm for Active Control of Civil Structures

The objective of this study is to formulate and assess a simple time-domain control algorithm. The algorithm under consideration applies a control force to the structure at the current time step in order to cancel the portion of the system velocity that is induced by the response and the external load at the preceding time step. The effectiveness of the control algorithm is quantified using representative SDOF and MDOF test structures subjected to several earthquake time histories. The performance of the time-domain control algorithm is comparable with the performance of the classic linear quadratic regulator (LQR) formulation. Furthermore, the time-domain algorithm seems to be almost insensitive to uncertainties in the structural parameters and is able to accommodate practical limitations, such as maximum actuator capacity, without a drastic degradation in performance. Lastly, for MDOF structures, it is possible to develop and compare several control schemes depending on which velocities are selected to be independent of the response and loading at the previous time step.

Adaptive Design Models Considering Lifespan of Buildings

A new design procedure is proposed which integrates system identification procedures and fuzzy set mathematics with damage and degradation models to formulate an adaptive design model capable of considering the changes a structure experiences dur ing its lifespan. Using this design model, the original finite element model can adapt to the assumptions inherent in finite element analysis and to the structural degradation that takes place as the system ages.

Efficient Integration of the Time Varying Closed-Loop Optimal Control Problem

Much of the research done in recent years towards the development of the smart or adaptive structure focuses on the application of active control to alleviate undesirable structural responses. However, classical optimal control algorithms are not directly applicable to most structural engineering applications because the control gains, obtained by solving the matrix differential Riccati equation (DRE), neglect the effects of the external forcing function and assume a time invariant system (which allows the DRE to be reduced to an algebraic Riccati equation). Due to these simplifications, the control algorithm loses its optimality and consequently may induce significant control inefficiency. In this study, a matrix-valued integration procedure is formulated for and applied to the differential Riccati equation with time variant plant and weighting matrices. The ability to effectively integrate the time variant differential Riccati equation allows the control algorithm to adapt to changing structural parameters. Examples are presented which apply this new procedure to solve the control equations associated with a five story, tendon-controlled shear building with time variant stiffness and damping matrices.

Dual forms and a proposed forward integration method for the matrix differential Riccati equation

Much of the research done in recent years towards the development of the smart or adaptive structure focuses on the application of active control to alleviate undesirable structural responses. Classical optimal control algorithms are not directly applicable to most civil engineering applications because the control gains neglect the effects of the external forcing function and assume a time invariant system (which allows the differential Riccati equation (DRE) to be reduced to an algebraic Riccati equation). The reason for these assumptions is that the time dependent DRE can only be stably integrated backwards in time. This study presents a more effective LQR control algorithm for civil structure applications, based on a proposed methodology for forward integrating the DRE. A set of dual equations are presented together with an optimization technique for obtaining the DRE solution from a forward integrable dual form. In addition, a matrix-valued integration procedure is formulated for and applied to the differential Riccati equation with time variant plant and weighting matrices.

An advance notification system for smart structures in seismic zones

This paper presents an advance notification system which links smart structures in seismic regions and enables them to learn of impending earthquakes prior to the initial excitation. When an event occurs, the smart structure or sensor nearest the epicenter is triggered by the initial P wave and calculations are performed to estimate the earthquakes predominant frequency, epicentral distance, and magnitude. This information then is sent to the remaining smart structures via a networked telecommunications system where it is used to prepare active control systems and expedite emergency management procedures.