J.M. Maciejowski. Multivariable Feedback Design, Addison-Wesley Publishing, Wokingham, England, 1989.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....Not surprisingly, tremendous research effort subsequently went into the development of design techniques which were based on optimisation principles, but which allowed robustness properties to be built into the controller directly. The result of this effort is a comprehensive theory ( 45] [60], 103] which has its origins in the seminal paper of Zames [99] and which has come to be known as control theory. Central to this approach is the interconnection structure shown in Figure 1.1, which shows the interconnection of a finite dimensional linear time invariant plant P, with a ....

....standard 72 contour 1. We next state some important results concerning the factorisation of transfer functions of finitedimensional LTI systems over the ring of matrices with elements in RHo. Such factorisations have played a central role in the development of multivariable control theory ( 45] [60], 84] 103] providing useful characterisations of dosed loop stability, a transparent parameterisation of sta bilising feedback schemes and a powerful tool in approximation (model reduction) theory. The transfer function of a finite dimensional system can always be written in terms of a quotient ....

[Article contains additional citation context not shown here]

J.M. Maciejowski, Multivariable Feedback Design, Wokingham, Addison Wesley, 1989.

....of worst case resource requirements can result in an extremely expensive and underutilized system. As a cost effective approach to achieve performance guarantees in unpredictable environments, several adaptive scheduling algorithms have been recently developed (e.g. 5] 8] 9] 24] 44] 46][55]) While early research on real time scheduling was concerned with guaranteeing complete avoidance of undesirable effects such as overload and deadline misses, adaptive real time systems are designed to handle such effects dynamically. There remain many open research questions in adaptive ....

....rather than systematic control derivation to achieve performance guarantees. In recent years, QoS adaptation architectures and algorithms have been developed to support applications such as communication subsystems [8] multimedia [19] 24] distributed visual tracking [46] and operating systems [55][61] 63] 69] 78] Some of these techniques [55] 61] 63] include optimization algorithms to optimize the value in QoS adaptation. However, their optimization algorithms assume that the resource requirement of every QoS level is a priori known. In contrast, our FCS framework provides performance ....

[Article contains additional citation context not shown here]

J. M. Maciejowski, Multivariable Feedback Design, Addison-Wesley, 1989.

.... H 2 design methods for local vehicle tilt control. The controller structures designed use two measurements: i) body lateral acceleration, ii) secondary suspension roll position, i.e. exactly the same measurements as used in both the nulling strategies. For the purposes of control [4, 8], the design objectives must be formulated as an optimisation problem, defined in the generalised regulator setting shown in Figure 5(b) In this diagram, P (s) represents the generalised plant , consisting of the nominal model G(s) combined with all frequency weightings appropriately chosen to ....

Maciejowski, J. M., Multivariable feedback design, Addison-Wesley, 1989.

....to tune. Our algorithm can be used to provide an adequate QFT design, or as the first step for designing a more complcx controller [4] In addition, the algorithm can easily tackle a large number of constraints and can be also be applied to multivariable systems, using the standard QFT approach [5, 6]. A PID controller has a transfer function gpld(s) p q ki q. d S (3) S and is therefore completely defined by the three terms kp (proportional gain) ki (integral gain) and ka (derivative gain) Its frequency response is Kpid(J60) kp J q JkdOO (4) Suppose that Kpd(s) is used in ....

MACIEJOWSKI, J.M.: 'Multivariable feedback design' (AddisonWesley,

....influence, and output influence matrices, respectively. The vectors w and v are the process and measurement noise vectors, respectively, and are modeled as zero mean and uncorrelated white noises with covariances ww W 0 ; E vv = V 0 (8) The optimal LQG controller is given by [3] x = A 0K f BK c )x K f y u = c x (9) K c = 0R P c ; K f = P f C V 01 (10) where P f and P c are positive semidefinite matrices that satisfy the following algebraic Riccati equations. P c P c A0 P c BR 01 B P c C QC = 0 P f A AP f P f C V 01 CP f ....

.... AP f P f C V 01 CP f 0 W0 = 0 (11) 3 If the triples 0 A; B; Q 1=2 C A; 0W ; C are stabilizable and detectable, positive semidefinite solutions to the Riccati equations in eq. 11) exist and the resulting controller is stabilizing, i.e. the closed loop system is stable [3]. Loop transfer recovery is a way of designing the LQG controller such that the desirable robustness and performance characteristics of the full state feedback are recovered at the plant input or output. For a square plant, a two step procedure is followed to achieve LTR at the plant output [3] ....

[Article contains additional citation context not shown here]

J. M. MacieJowski, "Multivariable Feedback design," Addison-Wesley Publishing Co., New York, NY, 1989.

....that an eigenlocus # i (##) does not form a complete closed graph in the complex plane. It can be shown that in this case all eigenloci drawn together in one diagram form a number of complete closed curves. The graphical condition then has to be applied to these completed loci, for details see [Mac89]. We will find that in the RNN case the eigenloci are circles or from circles derivable well defined closed curves in the plane. 5. The (multivariable) Nyquist theorem can be stated for more general operators including instable elements in the feedforward path. We do not need these more general ....

....the feedforward path. We do not need these more general For the example this has been done with the help of Maple V, Release 4 concepts, because in the RNN case we always assume that G is a stable linear operator. 6. Our treatment here is an appropriate mixture of different criteria given in [Mac89], Theorem 2.8, Vid93] Theorem 6.5.35, and [DV75] Theorem 4.36. We need the multivariable Nyquist theorem to assure stability of the RNN loop transformed operator G # shown in Fig. 4.4) We evaluate the condition (4.20) for two typical G # considered in the next examples. Example 17. Let G(s) ....

J. M. Maciejowski. Multivariable feedback design. Addison Wesley, 1989. 4.3, 4.3

....follows that the sensitivity transfer matrix S is small and T I. Similarly, if all the singular values of the loop transfer matrix are small, then the loop transfer matrix is small in all directions, and it follows that T is small and S I. These important ideas are discussed in, for example, [11, 6, 14, 19, 18, 4]. At in band frequencies, singular value loop shaping specifications have the form oe min (L(j ) l( 1; where l is some frequency dependent bound. For cutoff frequencies, singular value loop shaping specifications have the form oe max (L(j ) u( 1; where u is some frequency ....

J. M. Maciejowski. Multivariable Feedback Design. Addison-Wesley, 1989.

....single input single output to multi input multi output systems, the classical techniques became cumbersome, or even impossible to apply. A new representation was developed: the state space representation. An nth order differential equation is converted into n first order differential equations [Mac89] State space representation has various advantages such as simplicity and multiinput multi output representation. Techniques such as Optimal, Robust and Adaptive Control were developed. For linear systems, theory and algorithms for analysis and design of control systems are well understood. ....

....by Gaussian noise. This technique synthesizes a feedback control law which minimizes the average or expected cost of a quadratic function. The state and control variables are random variables. In the literature, other important control techniques such as Robust Control and can be found [Jac93,Mac89] 4.3 Nonlinear Control Systems Transient stability of power systems is modeled as systems of highly nonlinear differential equations (swing equations) Therefore, it is worth considering tools for analysis and design of nonlinear control systems. This subsection reviews basic literature in ....

J. M. Maciejowski. Multivariable Feedback Design. Addison-Wesley, 1989.

....presented a complete solution of the H1 control problem based on time domain techniques. This approach was based on Riccati equations and hence could be implemented easily. A first solution of the discrete time H1 problem can be found in [66, 49, 9] Currently several good books are available [10, 67, 50, 31]. The main reason for studying the H1 control problem is model uncertainty. Using H1 , it is possible to suppress the effect of model uncertainty on the behaviour of our plant. Also dependence on our knowledge of noise characteristics can be handled via H1 control. We consider the following ....

J.M. Maciejowski, Multivariable feedback design, Addison-Wesley, Reading, MA, 1989.

....results. 8 Conclusions The subject of this report are problems of Iterative Learning Control with nonminimum phase plants, i.e. plants with one (or more) zeros z in the right half plane. It is a commonplace fact that nonminimum phase plants have in classical control performance limitations [10]. This is due to the fact that their inverse is unbounded for infinite times. In Iterative Learning Control, only a finite time interval is used. While the inverse of the plant exists there, the analysis has shown that it has a very large norm. By splitting the input space into a part U which ....

J. M. Maciejowski. Multivariable Feedback Design. Addison-Wesley, Wokingham, 1989.

....scaling of the outputs is an important factor in determining the fundamental limitations that arise in the control of SITO systems. Keywords: Multivariable systems; Sensitivity Analysis; Inverted Pendulum 1 Introduction The control of multivariable systems has been the topic of extensive research(Maciejowski 1989). Much of this research has been based around the development of techniques to design a controller that optimizes a scalar cost function. Various techniques exist for the optimization of scalar cost functions including LQG techniques(Anderson Moore 1971) and H 1 optimization (Zhou, Doyle ....

Maciejowski, J. M. (1989), Multivariable Feedback Design, Addison-Wesley.

....def = LO (s) I LO (s) Gamma1 and T I (s) def = L I (s) I L I (s) Gamma1 . For a given plant G(s) we say that the closed loop is stable, and that the controller K(s) is stabilising, if each of the four closed loop transfer functions SO (s) S I (s) TO (s) and T I (s) is stable [20]. Each closed loop transfer function characterises a closed loop input or output response to one or more of the various exogenous disturbances. For example, the output sensitivity function may be interpreted as the transfer function from an output disturbance to the output, and the input ....

....as the transfer function from an output disturbance to the output, and the input complementary sensitivity function is the transfer function from an input disturbance to the input. A more comprehensive physical interpretation of each of these transfer functions may be found, for example, in [20, 10]. It is often desirable that these closed loop transfer functions be small, particularly in the frequency region(s) where the spectrum of the disturbances is significant. If absolutely nothing is known about the spectral content of the exogenous disturbances, a reasonable design goal might be to ....

J.M. Maciejowski. Multivariable Feedback Design. Addison Wesley, Wokingham, England, 1989.

....poles against the lower horizon # # . The actual poles are asymptotic to the desired closed loop poles at # = #5## = #4#33 # 2#50#. Numerical problems occur for # # # 5 again, this suggests choosing # # =10# # is a good rule of thumb. 4. 3 Turbogenerator system This example is taken from Maciejowski (1989) and represents a turbogenerator as a linear, twoinput, two output, six state system with the six open loop poles at: # = #0#234 (34) # = #0#349 # 6#34# (35) # = #1#04 (36) # = #10#5 (37) # = #15#9 (38) This system has two inputs. The basis functions of each input are chosen to be second order ....

Maciejowski, J. M. (1989). Multivariable Feedback Design. Addison-Wesley.

....matrices of the neural network model or neural controller. This leads to perturbed NL q systems, which correspond to nominal nonlinear models, with uncertainty modelled as a feedback perturbation upon this nominal model. Similar representations arise in modern robust control theory such as theory [5, 7], where uncertainties upon nominal linear models are modelled by means of a feedback perturbation through LFTs (Linear Fractional Transformations) It turns out that in the case of real parametric uncertainties the state space upper bound test in theory can be considered as a special case of the ....

.... = WAB oe(V A x k VB WG j k fi AB ) K ffl k z k 1 = WEF oe(V E z k V F WC k V F ffl k V F2 d k fi EF ) k 1 = oe(V C WAB oe(V A x k VBWG j k fi AB ) VC K ffl k ) j k 1 = oe(V GWEF oe(V E z k V F WCD k V F ffl k V F2 d k fi EF ) 6) which is in standard plant form [1, 5] an NL 2 system with state vector p k = x k ; z k ; k ; j k ] and exogenous input w k = d k ; ffl k ; 1] Also linear controllers or parametrizations by multilayer perceptrons with two hidden layer could be considered, leading to NL q systems for the closed loop system as well. 4 Perturbed NL q ....

[Article contains additional citation context not shown here]

Maciejowski J.M., Multivariable feedback design, Addison-Wesley, 1989.

....in this general area. In this paper the control objective considered is to minimize the displacement at a specified point on the panel in the presence of point force disturbances acting at other location(s) on the panel. The control strategy is based on LTR. The LTR design methodology followed [3] is the definition of a target feedback loop (TFL) which is then recovered through an asymptotic design. In particular, we follow the well known two step design procedure for recovery at the input of a square plant and its dual for recovery at the plant output. The actual designs were undertaken ....

J. M. Maciejowski, "Multivariable Feedback Design", Addison Wesley, 1989. 5 6

....of these points the system is linearized and a linear stabilizing controller is calculated for each specific operation point according to classical or modern linear control theory (e.g. by pole placement, PID, LQR, H 2 , H1 , etc. see e.g. Astrom, Wittenmark (1984) Franklin et al. 1990) Maciejowski (1989) for an introduction to linear control theory) A general parametrized control law is proposed for the switching from one operating point to another, either by a feedforward or a recurrent neural network depending on the linear controller design. This control law is overparametrized but the ....

....as will be illustrated for an inverted pendulum system in section 5. For the linear controller design on the linearized system all existing linear controller design methods can be applied, including e.g. PID, LQG, H 2 , H1 , synthesis etc ( Astrom, Wittenmark (1984) Franklin et al. 1990) Maciejowski (1989)) The ideas can also be extended to parametrizations with more than one hidden layer like in the static output feedback case. 6 4 Transition between equilibrium points Transition between equilibrium points will be discussed here for the static and dynamic output feedback case. 4.1 Static ....

[Article contains additional citation context not shown here]

Maciejowski J.M. (1989). Multivariable feedback design, Addison-Wesley.

....that neural networks can do potentially more than linear controllers: the neural controller can be designed such that it behaves as a linear controller (ranging from classical control techniques such as PID, pole placement to modern linear controller design as H 2 , H1 , synthesis etc. [8]) around the target equilibrium point but with the additional capability of realizing a transition between a point outside the region where this linearization is valid to the target equilibrium point. This transition is formulated as a nonlinear optimization problem with the linear controller ....

Maciejowski J.M., Multivariable feedback design, Addison-Wesley, 1989.

....=#I CP# ,1 #D x CN x # and with P x = N x D ,1 x this yields D =#I CP# ,1 #I CP x #D x . N follows from N = PD. 4. IDENTIFICATION AND CONTROL The general feedback matrix T #P;C# has been recognized as an important feedback property of the closed loop system #Bongers and Bosgra, 1990; Maciejowski, 1989#. It induces a feedback relevant topology, see also Schrama #1992b#, meaning that if two such operators are alike, the corresponding feedback controlled systems will have similar performances. Moreover, the control design being used in this paper is based on the minimization of the 1 norm of the ....

....of the plant P . However #19# introduces an additional parametrization constraint on the rcf # N; D# of the model P that has to be taken into account while performing an identi#cation of the nominal model P . D#q;## CN#q;##=D x CN x #20# By replacing the norm operator k#kby the H 2 #norm #Maciejowski, 1989# the following quadratic feedback relevant performance criterion J f ###, based on #17#, can be de#ned, if the parametrization constraint given in #20# is satis#ed 2 for notional convenience, the parameter # being estimated will be omitted without mentioning J f ### def = 1 4# Z # ,# ....

Maciejowski, J.M. #1989#. Multivariable Feedback Design. Addison#Wesley Publishing Company,Wokingham, UK.

....r 1 , r 2 to outputs of interest e 1 , e 2 . The above form of the performance can physically represent regulation, tracking, and or disturbance rejection problems. In general, any suitable multivariable design approach can be used to generate a substructure robust controller. References [7, 19] describe several control design approach including parameter optimization via nonlinear programming. In this paper, we consider the robust performance measure in terms of the structured singular value. The problem then reduces to minimizing (see for example [20, 21, 22, 23] 4 Substructure ....

Maciejowski, J.M., Multivariable Feedback Design, Addison-Wesley Publishing Company, Reading, MA, 1989.

.... ) a pre compensator structure and some frequencies together with their weights are selected and the diagonalization is achieved at these frequencies through optimization of a suitable cost function and nally the other frequencies on G s are checked (Hawkins 1972, Sain 1978, Ford and Daly 1979, Maciejowski 1989, Karimi 1995, Karimi and Tahani 1996) The proper choice of the number of frequencies, their values and their weights in the cost function demands a very large number of trial and error computations and sometimes will not be accomplished successfully (Maciejowski 1989) We introduce a new method ....

.... Sain 1978, Ford and Daly 1979, Maciejowski 1989, Karimi 1995, Karimi and Tahani 1996) The proper choice of the number of frequencies, their values and their weights in the cost function demands a very large number of trial and error computations and sometimes will not be accomplished successfully (Maciejowski 1989). We introduce a new method to achieve pseudo diagonalization through optimization of an interesting function. The presented method shows remarkable advantages over other existing methods. The rst one is that the number of frequencies used in the cost function is o# ered by interpolation the ....

[Article contains additional citation context not shown here]

Maciejowski, J. M., 1989, Multivariable Feedback Design (Addison Wesley).

....uncertainties are known to be structured and or parametric, and a characterization based on the H norm then leads to a conservative design. Robustness with respect to structured norm bounded uncertainties can be represented in terms of the structured singular value (Doyle 1982, Doyle et al. 1982, Maciejowski 1989, Skogestad and Postlethwaite 1996) In this approach the uncertainty is described in terms of a structured uncertainty block. Worst case (H ) performance and structured uncertainties can be treated in a unified framework using the synthesis technique (Doyle et al. 1982, Maciejowski 1989, ....

....1982, Maciejowski 1989, Skogestad and Postlethwaite 1996) In this approach the uncertainty is described in terms of a structured uncertainty block. Worst case (H ) performance and structured uncertainties can be treated in a unified framework using the synthesis technique (Doyle et al. 1982, Maciejowski 1989, Skogestad and Postlethwaite 1996) It is, however, very hard to design an H 2 optimal controller which achieves robustness against structured uncertainties described by a structured uncertainty block. Also, the available information about process uncertainties is often in a form which is not ....

[Article contains additional citation context not shown here]

Maciejowski J.M (1989). Multivariable Feedback Design. Addison Wesley Publishing Company, Inc.

.... closed loop dynamics is obtained by further manipulation: 2 I QA 01 1 A 3 3 l = QA 01 1 [A 2 l 0 A 4 l e 0 f e0 ] 30) where, A 3 (s) 0 J T M e J MR 1 s 2 J T B e Js J T K e J (31) A 4 (s) J T B e Js J T K e J (32) The multivariable Nyquist criterion (see [14]) can now be used to determine system stability. 5. EXPERIMENTAL RESULTS This section reports the single cylinder experiments that were carried out using the stick actuator of the mini excavator. For this purpose, the stick piston was disconnected from the rest of the machine (to eliminate the ....

J.M. Maciejowski, Multivariable Feedback Design. Addison-Wesley, 1989.

....demanding, no controller can be found. In this case, the objectives have to be adjusted. An efficient synthesis method has been developed using state space algorithms. However, this approach may lead to conservative results, since it neglects possible knowledge of structure in the uncertain model [4, 17]. A way to gain better results is to exploit available information about the structure of the uncertainty. Throughout the system, uncertainties may be present. Instead of modelling these uncertainties as unstructured, the general uncertainties at the various components can be taken into account ....

....of the new standard plant. 2. 9 Additional Reading An extended version of this chapter can be found in Goverde [14] Most considered concepts are also covered in Balas et al. 1] The state space approach to H1 optimal control can be found in Doyle et al. 4] Glover and Doyle [13] Maciejowski [17], and Safonov et al. 19] A reference to the theory of LFTs is Doyle et al. 5] The use of LFTs with respect to real parameter variation can be found in Lambrechts et al. 16] and Terlouw and Lambrechts [21] Computational issues of are given in Fan and Tits [6] Fan et al. 7] and Young et ....

Maciejowski, J.M., Multivariable Feedback Design, Addison-Wesley, Wokingham, 1989.

.... us with a stopping criterion for the algorithms developed in [4, 5, 6] For the lower bound computation let us introduce the Youla parameterization u b y = J y b u ; b u = Qby; Q 2 RH1 2 of all internally stabilizing controllers, where J is an (easily computed) LTI system [7] which is chosen such that Q enters the controlled system affinely as C j = T j 1 T j 2 QT j 3 ; again with some (easily computed) T j 1 , T j 2 , T j 3 in RH1 . This leads us to the problem considered in this paper. Problem. Suppose we are given T j 1 , T j 2 , T j 3 in RH1 ....

J. Maciejowski, Multivariable Feedback Design. Addison-Wesley, Wokingham, England, 1989.

....mathematically sophisticated methodologies for designing controllers for certain well defined classes of control problems. For example, proven design techniques exist for control problems involving linear processes and certain types of objective functions, such as quadratic cost functions [86]. Using these techniques, 4 controllers can be designed that are provably stable, have desired response characteristics, or perform optimally. However, despite substantial efforts, similar techniques are not yet available for more general classes of control problems involving processes that are ....

Maciejowski, J. M. Multivariable Feedback Design. Addison Wesley, 1989.

....variables, the parameter transformation suggested in this section can save a considerable amount of computation time. 3 MO Control by Dynamic Output Feedback In a general interconnection structure the property (8) does typically not hold. It is, however, well known that the Youla parameterization [14] is the tool to enforce this condition. Indeed, let us choose K, L such that A BK, A LC are stable. Then the set of all stabilizing controllers for (1) can be parameterized as u b y = 2 6 4 A BK LC GammaL B K 0 I GammaC I 0 3 7 5 y b u ; b u = Qby; Q 2 RH1 : 7 The ....

J. Maciejowski, Multivariable Feedback Design. Addison-Wesley, Wokingham, England, 1989.

....Z t h tjt 1 will often be uncorrelated with h t ijt i 1 , but when Z t 6= 0 it will never be independent, due to the feedback loop in Figure 1. The loop could become unstable. As discussed in [8] the gain of L 1 (q 1 ) cannot be allowed to be arbitrarily large and the small gain theorem [12] will provide (conservative) sucient conditions for stability. In [8] three important scenarios are discussed in which an exact stability, convergence and performance analysis is possible assuming v t , t and e t to be mutually independent. e t H h t R j t v t f t L ....

J. M. Maciejowski, Multivariable Feedback Design. Addison Wesley, Reading, Mass, 1989.

....of the Gradient Noise The feedback noise Z t h tjt 1 will not be independent of h t ijt i 1 , due to the feedback loop in Figure 4 and the loop could become unstable. As discussed in Part II [26] the gain of L 1 (q 1 ) cannot be allowed to be arbitrarily large but the small gain theorem [30] will provide (conservative) sucient conditions for stability. Since the properties of t depend on L 1 , a tracking design will require a few iterations, as outlined in Section V. After each iteration, we may have to estimate the properties of t by simulation. However, in Part II [26] three ....

J. M. Maciejowski, Multivariable Feedback Design. Addison Wesley, Reading, Mass, 1989.

....of M(s) No solutions are missed because e D(s) e.g. d(s) is of full normal rank. Thus the problem of finding a minimal polynomial basis to NL (M(s) has been transformed into finding a minimal polynomial basis to NL ( f M 1 (s) 3 3 Design Example: Aircraft Dynamics This model, taken from (Maciejowski 1989), represents a linearized model of vertical plane dynamics of an aircraft. This section includes MATLAB code of central operations. The inputs and outputs of the model are Inputs Outputs u 1 : spoiler angle [tenth of a degree] y 1 : relative altitude [m] u 2 : forward acceleration [ms 2 ] y 2 ....

Maciejowski, J. (1989). Multivariable Feedback Design, Addison Wesley.

....the feasibility of the approach and demonstrate the benefits of the new developments. Introduction Robust control theory guarantees that a feedback control system can be designed that will maintain desired levels of stability and performance subject to modeling errors and uncertainties. [1,2] There is, however, an underlying assumption that the uncertainty model used in the design effectively characterizes the differences between the responses of the true system and the nominal design model. This means that the family of responses associated with the design model contain the ....

....conservatism. Robust Control Models Robust control design requires a linear design model that characterizes parameter variations, unmodeled dynamics, and or other uncertainties relative to a nominal model. The common format for describing such a model in block diagram form is shown in Figure 2. [1,2] The nominal model is represented by the (2,2) element of the transfer matrix P while the other elements represent various aspects of the uncertainty. The matrices W and D B characterize the structure of the uncertainty in the model; W is a real diagonal positive definite weighting matrix and D B ....

[Article contains additional citation context not shown here]

Maciejowski, J.M.: Multivariable Feedback Design. Addison-Wesley Publishers, Ltd. 1989.

....significant uncertainties. Therefore it is important to investigate the robustness of the controllers in the presence of parametric uncertainties in the design model. Robustness to structured parametric uncertainties can be investigated using the structured singular value (or Gamma) analysis [13]. The problem with standard Gammaanalysis is that it assumes the uncertainties to be complex valued rather than real which usually results in overly conservative estimates of permissible parametric uncertainty. The development of Gammaanalysis and synthesis methods for real parametric ....

Maciejowski, J. M.: Multivariable Feedback Design, Workingham, U.K.: AddisonWesley, 1989.

....is easy to understand, it has poor numerical properties. However the algorithm based on polynomial echelon form is both fast and numerically stable and should therefore be the preferred choice. 1 In version 1.6, the command is xab. 5 Design Example: Aircraft Dynamics This model, taken from [7], represents a linearized model of vertical plane dynamics of an aircraft. The inputs and outputs of the model are Inputs u 1 : spoiler angle [tenth of a degree] u 2 : forward acceleration [ms 2 ] u 3 : elevator angle [degrees] Outputs y 1 : relative altitude [m] y 2 : forward speed [ms 1 ....

J.M. Maciejowski. Multivariable Feedback Design. Addison Wesley, 1989.

.... Numerous simulations of linear MIMO CGPC are presented by Demircioglu and Gawthrop (1992) here, we concentrate on the differences arising from the use of the p(s) polynomial, which was not discussed by Demircioglu and Gawthrop (1992) A number of multivariable benchmark systems are given by Maciejowski (1989). GPC controllers have been designed for each of these and simulated using Matlab and the tool box associated with this paper. Rather than present numerous Figures, the design and results for just the three input, three output 5 state aircraft model (AIRC) are presented here. The other examples ....

Maciejowski, J. M. (1989). Multivariable Feedback Design. AddisonWesley.

....as the block LMS algorithm of Ferrara [24] and others. 3.3 FIR Matrices: FIR Polynomial Matrices and FIR Filter Matrices 3.3. 1 History and Introduction Polynomial matrices are discussed in the geophysical signal processing literature [20, 67] and in the controls and linear algebra literature [68, 79, 44, 60]. The geophysical literature describes polynomials with matrix coefficients (as in the Cayley Hamilton Theorem A matrix satisfies its own characteristic polynomial) The controls literature polynomial matrices are defined using a rational function (or ARMA) model. This group of polynomial ....

J. M. Maciejowski, Multivariable Feedback Design, Addison-Wesley, 1989.

....to demonstrate the ideas. Many problems in control theory seem to fit into the framework described in this paper. Some further examples of areas in control theory where applications of quantifier elimination methods could be investigated are: i) Feedback design of linear dynamical systems, see Maciejowski (1989). Stability and performance constraints are often given as semi algebraic constraints on the so called Nyquist curve. ii) Stability analysis using the circle and Popov criterion, see Vidyasagar (1993) iii) Computation of robustness regions of nonlinear state feedback, see Glad (1987) iv) ....

Maciejowski, J. (1989). Multivariable Feedback Design. Addison-Wesley.

....the observability canonical form has the least favorable sensitivity of the four different realizations. 2 9 Examples In this section we will present three examples illustrating the previously discussed properties of the proposed identification algorithm and model structure. Example 9. 1 In [21] Appendix A.2 a turbo generator model with two inputs, two outputs and six states is presented. We used the continuous time model to generate an estimation data set and a validation set using random binary ( Gamma1; 1) signals as inputs. The sample time was set to 0.05. The Model FM IM 6 = ....

J. M. Maciejowski. Multivariable Feedback Design. Addison-Wesley, Wokingham, England, 1989.

....and the sufficient condition for internal stability of T (M ; Delta) becomes sup oefM(e i )g = kMk1 fl Gamma1 . 2 3.2.2 Necessary condition The inequality in theorem 3. 2 is a sufficient condition for internal stability of the basic perturbation model T (M ; Delta) It can be shown (Maciejowski, 1989) that it is possible to find perturbations that violate theorem 3.2, but do not destabilize T (M ; Delta) However, if all perturbations can occur for which the bound (10) holds, theorem 3.2 becomes sufficient as well as a necessary condition for internal stability of T (M ; Delta) Theorem 3.3 ....

....as a necessary condition for internal stability of T (M ; Delta) Theorem 3.3 Let M , Delta 2 IRH1 with k Deltak 1 fl. The basic perturbation model T (M ; Delta) is internally stable 8 k Deltak 1 fl if and only if kMk1 fl Gamma1 Proof: Proven in theorem 3. 2 ( Has been worked out in Maciejowski (1989) or Doyle et al. 1992) and is based on proving the existence of a destabilizing perturbation if kMk1 = fl Gamma1 . 13 In cases that Delta has some special structure, e.g. a block diagonal structure, theorem 3.3 is no longer a necessary but only a sufficient condition. This is due to ....

[Article contains additional citation context not shown here]

Maciejowski, J.M. (1989). Multivariable Feedback Design. Addison--Wesley Publishing Company, Wokingham, UK.

....LQG LTR procedure, is similar to the chain for LQG design discussed in section 4.5. The difference is that in LQG LTR the cost weights Q and R, and the covariance matrices W and V are iterated in a particular way on the basis of computed frequency response properties. For details see chapter 5 of [57]. In particular, a common choice for the cost function is to set G = B and Q = C T C. The Container cost therefore becomes a Method Container, rather than a Data Container, as shown in figure 5.3, since it obtains B and C from the Container predictor. cost rgltr 10 2 1 2 freq estmtr ....

J.M. Maciejowski. Multivariable Feedback Design. Addison-Wesley Publishers, 1989.

No context found.

J.M. Maciejowski. Multivariable Feedback Design, Addison-Wesley Publishing, Wokingham, England, 1989.

No context found.

Maciejowski J.M., Multivariable feedback design, Addison-Wesley, 1989.

No context found.

Maciejowski J.M., Multivariable feedback design, Addison-Wesley, 1989.

No context found.

Maciejowski J.M., Multivariable feedback design, AddisonWesley, 1989.

No context found.

Maciejowski, J.M., Multivariable Feedback Design , Addison-Wesley 1989.

No context found.

J.M. Maciejowski. Multivariable Feedback Design. Addison Wesley, 1989.

No context found.

Maciejowski, J. M., Multivariable Feedback Design, Addison-Wesley Publishers Ltd, Wokingham, England, 1989.

No context found.

Maciejowski, J.M., Multivariable Feedback Design, 1989, Addison-Wesley Pub. Co., Wokingham, England.

No context found.

Maciejowski, J.M., Multivariable Feedback Design, Addison Wesley Publishing Co., Wokingham, England, 1989, Chapter 6.

No context found.

Maciejowski, J.M., Multivariable Feedback Design, Addison-Wesley Publishing, 1989.

No context found.

J.M. Maciejowski. Multivariable Feedback Design. Addison-Wesley Publishers, 1989.

No context found.

Maciejowski,J. M., 1989 Multivariable Feedback Design, Addison Wesley, New York.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC