Data-based uncertainty modeling of MIMO systems

Hamed Jafarian, Kurt Erik Häggblom

    Research output: Chapter in Book/Conference proceedingConference contributionScientific


    Many robust control design methods require a linear model consisting of nominal model augmented by an uncertainty description. A general form for such a model is

    (1)     G = G0 + H21Δ(I-H11Δ)-1H12

    where G is the transfer function of the true system, G0 is a nominal model, and Δ is a perturbation causing uncertainty about the true system. Depending on the particular type of uncertainty (additive, input or output multiplicative, inverse types of uncertainty, combinations of various types of uncertainty), H11, H12 and H21 can contain combinations of (known) constant matrices and the nominal model.

    Assume that we have information about the true system in the form of a number of possible transfer functions Gk, k = 1,...,N. The nominal model and the perturbation Δk associated with Gk are unknown, but they have to satisfy

    (2)     Gk = G0 + H21Δk(I-H11Δk)-1H12 ,   k = 1,...,N.

    How should G0 be determined? It has been shown that ||Δ|| is a control relevant measure of the distance between G and G0 for models of the form (1) and that the achievable stability margin by feedback control is inversely proportional to this distance. For a given type of uncertainty model, this suggests that G0 should be determined by solving the optimization problem

    (3)     minG0 maxk ||Δk||

    subject to the appropriate data matching condition (2). Obviously, the type of uncertainty model giving the smallest minimum is the best one according to this measure.

    If information about the system is obtain through identification, input-output data are available. An attractive way of removing noise from the output is to fit a model Gk to the data and to calculate a noise-free output yk by yk = Guk, where uk is the input in experiment k. Since the purpose of the experiments in this context is to excite the system in various ways, the inputs do not tend to be persistently exciting in all individual experiments. Thus, Gk only applies to the particular input uk, and the relevant information is input-output data {uk, yk}, k = 1,...,N. This means that the model matching condition (2) should be replaced by the input-output matching condition

    (4)     yk = G0uk + H21Δk(I-H11Δk)-1H12uk .

    It can be shown that the use of (4) instead of (2) results in a less conservative uncertainty model.

    Our modeling approach is to model G0 in the frequency domain using sampled frequency responses of the input-output data. Because of the availability of Gk, these are easy to calculate for standardized inputs. The available information is thus {uk(jω), yk(jω), ω∈Ω}, k = 1,...,N, and we solve the optimization problem frequency-by-frequency, i.e.

    (5)     minG0(jω) maxk ||Δk(jω)||2   s.t. (4), ∀ω∈Ω.

    The uncertainty Δk is assumed to be unstructured.

    For some types of uncertainty, the optimization problem can easily be formulated as a convex optimization problem. Additive uncertainty, for example, which in its basic form is described by

    (6)     yk = G0uk + Δkuk , k = 1,...,N,

    results in the optimization problem

    (7)     minG0γ  s.t. matrix(γI, yk - G0uk; (yk - G0uk)*, uk*uk) >= 0, k = 1,...,N,ω∈Ω

    which is a convex optimization problem. Here A* denotes the complex conjugate transpose of A and P>=0 denotes that P is positive semidefinite.

    Many types of uncertainty descriptions do not readily give a convex optimization problem. For example, an output multiplicative uncertainty described by

    (8)     yk = G0uk + ΔkG0uk , k = 1,...,N,

    results in the optimization problem

    (9)     minG0γ  s.t. matrix(γI, yk - G0uk; (yk - G0uk)*, uk*G0*G0uk) >= 0, k = 1,...,N,ω∈Ω

    which is non-convex due to the appearance of G0*G0. An iterative solution by keeping G0*G0 fixed during each iteration tends to produce local minima which are non-global. However, we show how this optimization problem, and similar ones for some other types of uncertainty, can be reformulated as a convex optimization problem.

    For control design, G0 is needed as a transfer function or a state-space model. In principle, we should determine such a model by replacing G0 in the appropriate consistency relations, like those appearing in (7) and (9), by a suitable parameterization of G0 . However, so far we have not been able to obtain a satisfactory solution in that way. Instead, we have fitted a model to the calculated frequency responses G0(jω), ω∈Ω. A drawback of this approach is that min||Δ(jω)||2 will increase, usually also min||Δ||, sometimes even drastically. We have studied various approaches of reducing the effects of this drawback.

    Original languageUndefined/Unknown
    Title of host publicationPreprints 16th Nordic Process Control Workshop
    EditorsTore Hägglund
    PublisherLund university
    Publication statusPublished - 2010
    MoE publication typeB3 Non-refereed article in conference proceedings
    Eventconference -
    Duration: 1 Jan 2010 → …


    Period01/01/10 → …


    • Convex optimization
    • Distillation columns
    • LFT uncertainty
    • Linear matrix inequalities
    • Linear multivariable systems
    • Robust control
    • Uncertainty modeling

    Cite this