Simulation and Modelling of Smart Beams with Robust Control Subjected to Wind Induced Vibration

This paper presents some recent developments in modelling and numerical analysis of piezoelectric systems and controlled smart structures based on a finite element formulation with embedded control. The control aims at vibration suppression of the structure subjected to external disturbances, like wind and noise, under the presence of model inaccuracies, using the available measurements and controls. A smart structure under dynamic loads is analysed and comparison between results for beam with and without control is made. The numerical results show that the control strategy is very effective and suppresses the vibrations of the structure.


Introduction
The intelligent material sector has broadly motivated research interest during the last decades. Materials are classified as intelligent if they exhibit the ability to sense a certain stimulus and initiate a reaction to it in a controlled manner [1][2][3][4][5][6][7][8][9][10][11]. An intelligent structure is a structural system or member with embedded sensors and actuators, combined with a control system that automates the structure's reaction to external loadings acting on it, by attenuating their influence or improving desirable responses. Development in this scientific discipline is aided by advances in material science as well as in the field of control. Material science research has spurred the development of new materials that can be incorporated in structures to be used as actuators and sensors [9][10][11].
In this paper, we address the problem of vibrations of intelligent structures. Stimuli may come from external perturbations of the system, disturbances or excitation that may cause structural vibrations, such as wind loading or earthquakes. An intelligent structure is expected to be able to sense the vibration and counteract it in a controlled fashion, so that vibration of the system be reduced and contained. To that end, a number of intelligent materials may be used as actuators and sensors. [9][10][11] Piezoelectric materials, memory materials, electrostrictive and magnetostrictive materials are such materials. In this work, we focus on the use of piezoelectric materials, given that they exhibit good sensing and actuation properties.
Among the commonly used optimal control schemes are LQR, and H  . It is known that if the controller is not robust enough, the uncertainties of the system may destroy the efficiency of the controller. The H  , control provides better robustness than LQR [12][13][14]. The aim of this work is to design an H  , robust controller for a beam bonded with piezoelectric sensors and actuators and to investigate the behaviour of the controlled beam. First, a detailed shear-deformable (Timoshenko) model for a laminated beam structure is developed. A finite element formulation is presented for the model. Quadratic Hermitian polynomials are used for the transverse and rotational displacements, respectively. The differential equations are based on the Timoshenko beam theory [15]. The governing state equation is established and used for the design of the control. The numerical simulations carried out on the laminated beam shows that the vibration of the system is significantly suppressed within the permitted actuator voltages. Herein the integration of control into a home-made finite element code developed in MATLAB is presented [16]. The numerical solution of the H  feedback controller has been done by using a nonconvex, non differentiable optimization approach with the usage of HIFOO software within MATLAB [14,17]. For the numerical results wind type loads are taken into account. The effectiveness of the technique in the modelling of the standard uncertainties is also presented.
tions is considered. Four pairs of piezoelectric patches are embedded symmetrically at the top and the bottom surfaces of the beam, as shown in Figure 1. The beam is made from graphite-epoxy T300-976 and the piezoelectric patches are PZT G1195N. The top patches act like sensors and the bottom like actuators. The resulting composite beam is modelled by means of the classical laminated technical theory of bending. Let us assume that the mechanical properties of both the piezoelectric material and the host beam are independent in time. The thermal effects are considered to be negligible as well [5,6]. The beam has length L, width W and thickness h. The sensors and the actuators have width b S and b A and thickness h S and h A , respectively. The electromechanical parameters of the beam are given in the Table 1.

Reduced Piezoelectric Equations
In order to derive the basic equations for piezoelectric sensors and actuators (S/As), we assume that:  The piezoelectric S/A are bonded perfectly on the host beam;   [5]. Under these assumptions the three-dimensional linear constitutive equations are given by [4], 11 31 where xx  , xz  denote the axial and shear stress components, z D , denotes the transverse electrical displacement; xx  and xz  are a axial and shear strain components; Q 11 , and Q 55 , denote elastic constants; d 31 , and ξ 33 , denote piezoelectric and dielectric constants, respectively. Equation (1) describes the inverse piezoelectric effect and Equation (2) describes the direct piezoelectric effect. E z , is the transverse component of the electric field that is assumed to be constant for the piezoelectric layers and its components in xy-plain are supposed to vanish. If no electric field is applied in the sensor layer, the direct piezoelectric Equation (2) gets the form, and it is used to calculate the output charge created by the strains in the beam [3].

Equations of Motion
The length, width and thickness of the host beam are denoted by L, b and h, respectively. The thickness of the sensor and actuator is denoted by h s and h a . We assume that:  The beam centroidal and elastic axis coincides with the x-coordinate axis so that no bending-torsion coupling appears;  The axial vibration of the host beam centreline is considered negligible;  The displacement field {u} = (ux, uy, uz) is obtained based on the usual Timoshenko assumptions, where φ is the rotation of the beams cross-section about the positive y-axis and w is the transverse displacement of a point of the centroidal axis (y = z = 0). The strain displacement relations can be applied to Equation (4) to give, We suppose that the transverse shear deformation is equal to zero [2]. In order to derive the equations of the motion of the beam we use Hamilton's principle, where T [6] is the total kinetic energy of the system, U is the potential (strain) energy and W is the virtual work done by the external mechanical and electrical loads and moments. The first variation of the kinetic energy is given by, The first variation of the kinetic energy is given by, If the load consists only of moments induced by piezoelectric actuators and since the structure has no bending twisting couple then the first variation of the work has the form [6], where M A is the moment per unit length induced by the actuator layer and is given by,

Finite Element Formulation
We consider a beam element of length L e , which has two mechanical degrees of freedom at each node: one translational ω 1 (respectively ω 2 ) in direction z and one rotational φ 1 (respectively φ 2 ), as it is shown in Figure 2.
The vector of nodal displacements and rotations q e is defined as [3], The beam element transverse deflection ω(x, t) and the beam element rotation ψ(x, t) of the beam are continuous and they are interpolated within by Hermitian linear shape functions i H  and i H  as follows [4], This classical finite element procedure leads to the approximate (discretized) variation problem. For a finite element the discrete differential equations are obtained by substituting the discretized expressions 12 into Equations (7) and (8) to evaluate the kinetic and strain energies. Integrating over spatial domains and using the Hamiltons principle 6 the equation of motion for a beam element are expressed in terms of nodal variable q as follows, where M is the generalized mass matrix, D the viscous damping matrix, K the generalized stiffness matrix, m the external loading vector and e the generalized control force vector produced by electromechanical coupling effects. The independent variable q(t) is composed of transversal deflections f f 1  and rotations 1  , i.e., [7]   1 1 n n q t where n is the number of nodes used in analysis. Vectors ω and m f are positive upwards. To transform to statespace control representation, let (in the usual manner), Copyright © 2012 SciRes. OJCE A. MOUTSOPOULOU ET AL.

109
The previous description of the dynamical system will be augmented with the output equation (some displacements or velocities are measured) [4], In this formulation u is n × 1 (at most, but can be smaller), while d is 2n × 1. The units used are compatible for instance m, rad, sec and N.

Statement of the Robust Control Problem
The optimal control problem is initially studied for the nominal system, i.e., the beam with known elastic, piezoelectric and viscous properties. A more realistic question concerning the robustness of the control in the presence of defects is also addressed. The fact that the system is in subjected to disturbances, such as the wind power, as well as the noise of measurements, is taken into account. The mathematical pattern being used in the design is an approximation of the real one. Further, two control laws for the composite beam are designed in order to suppress the vibrations. Because of its linearity and easy implementation, the linear quadratic regulator (LQR) [13] is presented first. The response of the controlled nominal and damaged beams is investigated. In order to take into account the incompleteness of the information about the eventual damages and external additional influences a robust H  controller is designed [12,14]. A system analysis is made on condition that the system is not accurate but includes uncertainty that may be related to some kind of damage [7].
For practical applications both algorithms need several trial-and-error design iterations in order to provide appropriate control voltages, since the piezoelectric actuators can be depolled by high oscillating voltages. The effectiveness of the proposed control strategies is investigated with the help of numerical simulations [3].

Robustness Analysis
The following three steps are taken in the robustness analysis [13,14,19]: 1) Expression of an uncertainty set by a mathematical model.
2) Robust stability (RS): check if the system remains stable for all plants within the uncertainty set.
3) Robust performance (RP): if the system is robustly stable, check whether performance specifications are met for all plants within the uncertainty set.
To perform the robustness analysis, the interconnection of Figures 3(a), (b) will be used.
Δ define the uncertainty, Μ define the nominal system, w are the inputs (the mechanical force and the noise of the system), z are the outputs (the state vector and the control vector). The uncertainty included in Δ satisfies 1    . Since controller K is known. Figure  3(a) can be simplified to Figure 3(b). Given this structure it is known that [16,19], where, is the structure singular value of M given the structured uncertainty set   . 2) The system (M, Δ) exhibits robust performance if, where, and   has the same structure as Δ but dimensions corresponding to (w,z). Unfortunately, only bounds on μ can be estimated.
To proceed let us assume uncertainty in the M, D and K matrices of the form, This means that we are allowing a percentage deviation from the nominal values [4]. With these definitions Equation (13) becomes, Two kinds of dynamic loading are used as disturbances: Writing (25) in state space form, gives,  Periodic sinusoidal loading pressure acting on the side of the structure simulating a strong wind. A sinusoidal load with an amplitude of 10 N and frequency of 6.2832 rad/sec, has been considered.

26)
 A typical stochastic wind-type load on the side of the structure (Figure 4).
In this way we treat uncertainty in the original matri s a cantilever composite

Results
The first load analyzed is sinusoidal load acting on the side of the structure. Figure 5 shows the dynamical response for the displacements of the uncontrolled and ces as an extra uncertainty term.

Numerical Examples
For the numerical simulation beam with viscous and piezoelectric layers bonded on its top and bottom and discretized with four finite elements, is used. A finer finite element discretization, which certainly is required for the approximation of higher frequencies, does not change the trend of the results. Our  olled beam with LQR control [12,13] and H  control [14,19], for the four nodes of the beam. Comparison with the open loop response for the same plant shows the good performance of the H  controller. Results are very good, and the beam remains in equilibrium. Reduction of vibrations is observed, while piezoelectric add-ons produce voltage within their tolerance limits (±500 volt).
Then a typical wind load (Figure 4) acting on the side of the structure. The wind load is a real life wind speed measurements in relevance with time that took place in Estavromenos of Heraklion Crete. We transform the wind speed in wind pressure with, where V = velocity, ρ = density and C u = 1.5. ponses are re In Figure 9 the structural displacement res ported, for the four nodes of the beam. It is possible to see clearly the benefit induced the control on the maximum value of these displacements. H  controller results are very satisfactory and prove that H  control can reduce smart structures vibrations. In Figure 10 we can see the structural rotations for the four nodes of the beam. The beam with H  control keeps in equilibrium and we have almost zero displacements.
In Figure 11 we can see the control profile for the four no

Reduced Order Control
∞ he fact that conlem is to compute a controller of redu des of the beams. As we can see the voltage is more less than 500 Volt, which is the piezoelectric limit.
The H controller found is of order 24. T troller order, which is equal to the order of the system, is relatively higher than the order of classical controllers such as LQR has led a number of researchers to develop order reduction algorithms. The most widely used such algorithm, known as HIFOO, has been implemented in a Matlab environment, and is the one used in the following procedure [17,20].
The general prob ced order n < 24 while retaining the performance of the H ∞ criterion as well as the behaviour of a full order n system [17,20]. Nonsmooth variation analysis and related computaan be effectively ap controller for the give tional methods are powerful tools that c plied to identify local minimizes of nonconvex +optimization problems arising in fixed-order controller design. Our computational methods found a 24 th order controller that stabilized the system. Using the Matlab package HIFOO we can reduce this controller and stabilize the system with a 2 nd order controller without diffi-  explicit formulas for the controller and r the closed loop system. Furthermore, our analytical practical interest, su culty, suggesting fo techniques prove that this controller is locally optimal in the sense that there is no nearby controller with the same order for which the closed loop system has all its poles further left in the complex plane [20].
These approaches can be extended in order to take into account other key quantities of great ch as optimization of H  performance. In particular, with the help of a MATLAB toolbox called HIFOO ( H  Fixed Order Optimization e can reduce the order of the controller and have very good results. Figure 12, show the response of the uncontrolled and control beam of the free end of the beam, using hifoo controller (a second order controller). Figure 13 shows the control voltages for the four nodes of the beam [20]. As we can see we need less energy then the 24 th controller for reduced vibrations.

Conc
) w s lusion rmulation and finite element model for n of a cantilever beam with piezo-A mathematical fo the vibration suppressio electric laminated surface and viscous layers and elastic core is presented in this paper. The design of the piezoelectric active control using LQR strategy and H  control theory for the nominal and damaged sandwich beam has been studied. The numerical results show at the th H  control strategy is very effective and suppresses the vibrations of the beam. The vector of active control forces