Jiann-Shiun Lew

Using transfer function parameter changes for damage detection of structures

A novel approach is presented for damage detection of large flexible structures by using the parameter change of the transfer function. First, an interval modeling technique, which represents the system uncertainty under the environmental change via the intervals of transfer function parameters, is used to distinguish the structural damage from the environmental change. In this paper a coherence approach is developed for locating the damage position when the structural damage is observed. Only a few sensors are required in using the proposed coherence approach for damage detection. This has the advantage of using the proposed algorithm for practical applications. Also this approach has the flexibility of using the multi-input and multi-output system. A nine-bay truss example is used to demonstrate and verify the approach developed. Nomenclature a, b = denominator and numerator parameters of transfer function C = coherence of the tested system to damage G = interval model g = transfer function k = number of modes m = number of outputs n = number of tests for environmental change p = parameter vector of transfer function R = magnitude ratio of the tested system to damage R, JR = magnitude ratio bounds W = weight of parameter change Wa, W}} = weights of intervals for interval model x,y = unit vectors of Cartesian coordinate system Ap = parameter change vector Superscripts

Structural Damage Detection Using Virtual Passive Controllers

This paper presents novel approaches for structural damage detection which uses the virtual passive controllers attached to structures. The passive control system which mimics the mass-spring-dashpot is an energy dissipation device augmenting the system damping and thus guarantees the closed-loop system stability. Two damage detection techniques are developed. One technique uses a direct output feedback controller whereas the other technique uses the second-order dynamic feedback controller. The change in the identified natural frequencies, which are generally much less sensitive to noise and environmental uncertainties than the identified mode shapes, are used for damage detection. A least-squares technique, which is based on the sensitivity of the natural frequencies to the damage variables, is used for accurately identifying the damage variables.

Interval model identification and robustness analysis for uncertain flexible structures

System identification and robustness verification techniques are implemented for vibration control of flexible structures. This control methodology allows engineers to develop models which account for parametric uncertainties in a system. The test bed is a 10-bay aluminum truss-like structure. A reaction mass actuator (RMA) is used as the force actuator device for controlling the structure. A local velocity feedback controller is utilized to suppress structural vibration. Added masses to structural nodes represent parametric uncertainties. Interval control system analysis was applied to show the limits of performance of the uncertain experimental system.

Robust control of identified reduced-interval transfer function

We propose and demonstrate the use of a reduced-interval transfer function for robust control design of a system with parametric uncertainty. A singular value decomposition technique is used to model a system with parametric uncertainty via an interval transfer function, where each interval represents one bounded uncertainty parameter. The interval lengths can be used to determine the number of intervals of the reduced-interval transfer function, which is used to form an uncertainty structure for application in robust control such as /spl mu/-synthesis. A major benefit of working with a reduced-interval model is the simplicity and computational efficiency in analyzing and designing the robust controller. Also the order of the designed controller may be significantly reduced by using a reduced-interval model. To verify the system performance of the designed controller based on the reduced-interval model, we present a combined algorithm consisting of a sensitivity analysis and interval polynomial techniques.

Quantification of Parametric Uncertainty via an Interval Model

The quantification of model uncertainty becomes increasingly important as robust control is an important tool for control system design and analysis. This paper presents an algorithm to characterize the model uncertainty in terms of parametric and nonparametric uncertainties directly from inputloutput data. We focus on the quantification of parametric uncertainty, which is represented as an interval system of the transfer function. Using this family of transfer functions (interval system), we give complete analysis of the system. A numerical example is used to demonstrate and verify the developed algorithm. The example illustrates the application of recently developed interval system techniques to the identified interval models.

Comparison of candidate methods to distinguish noise modes from system modes in structural identification

In modal identification, nonphysical noise or computation modes always appear to help match the input-output data. This paper studies the ability of four criteria to distinguish which modes in a model are noise modes: (1) modal amplitude coherency, (2) the relative contribution of each mode to the pulse response indicated by the mode singular value, (3) the variances of the mode frequencies and damping factors produced by a chosen measurement noise level, and (4) identification of the backward-time in order to let the shift from positive to negative damping of the true system modes distinguish these modes from noise modes. Both simulated and experimental data are used to study the four criteria.

Parametric Uncertainty Quantification of an Inflatable/Rigidizable Hexapod

Space exploration has always been constrained by our ability to develop systems that can be flown at reasonable cost within launch-vehicle volume constraints. Ultra-lightweight inflatable/rigidizable structures offer many advantages over conventional structures in this area. Recently, a 3m-diameter hexapod structure was designed and built, with these new materials and fabrication techniques, to conduct research on modeling and vibration control. For the space community to embrace this technology, systems like the hexapod must be studied to understand their performance. This paper presents a study of parametric uncertainty quantification of the dynamic model of this hexapod structure. Our goal is to develop a model with parametric uncertainty for robust control design and analysis. To obtain parametric uncertainty, experiments with various kinds of excitation, such as random input and sine-sweep input, at different levels of force are conducted. Time-domain and frequency-domain system identification techniques are applied to analyze the experimental data collected from the hexapod. A singular value decomposition technique is used to model the parametric uncertainty via an interval transfer function, where each interval represents one bounded uncertainty parameter.

Robust control of flexible structures against structural damage

This paper deals with the vibration suppression control problem of flexible structures against potential damage. The controller is required not only to robustly stabilize the structure but to satisfactorily suppress the vibration under all predicted damages that might occur in the 20-bay truss structure under study. The damage to the flexible structure we consider here is represented by the various stiffness losses of predetermined elements. The structure with the predicted damages is modeled as an interval transfer function. The controller design is based on the /spl mu/-synthesis in conjunction with the extremal properties of interval systems. Unlike conventional /spl mu/-synthesis, the technique enables one to use a fixed-size uncertainty block (4 real blocks for SISO case) regardless of the number of parameters, that results in a faster, reliable and simple synthesis procedure.

Existence and uniqueness proof for the minimum model error optimal estimation algorithm

A general solution to the jump-discontinuous two-point boundary value problem (TPBVP) of the minimum model error is presented. Its existence and uniqueness are then proved.

ROBUST CONTROL OF IDENTIFIED REDUCED-INTERVAL STATE SPACE MODEL

This paper proposes the use of an identified reducedinterval state space model for robust control design of a multivariable system with parametric uncertainty. First, an interval modeling technique is used to model a dynamic system with parametric uncertainty via an interval state space model, where each interval represents one bounded identified uncertainty parameter. The interval lengths indicate the distribution of the identified uncertainty parameters, and they can be used to determine the order of the reduced-interval state space model. This identified reduced-interval state space model is used to form the structural uncertainty for /^-synthesis for the robust control design. To verify the performance of the designed controller based on the reduced-interval model, we apply a sensitivity analysis. This sensitivity technique can also be used for model validation of the reduced-interval model. A 20bay truss structure with potential damage of various elements is used to demonstrate and verify the approach.