A decoupling pole placement self-tuning controller for a class of multivariable processes (Q3748920)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 3985081
Language Label Description Also known as
default for all languages
No label defined
    English
    A decoupling pole placement self-tuning controller for a class of multivariable processes
    scientific article; zbMATH DE number 3985081

      Statements

      A decoupling pole placement self-tuning controller for a class of multivariable processes (English)
      0 references
      0 references
      0 references
      1986
      0 references
      Pole-placement
      0 references
      multivariable self-tuning controller
      0 references
      decoupling
      0 references
      This paper describes a new multivariable self-tuning controller that specifically handles different time delays between each of the input- output pairs (multiples delays), and deals with unknown or varying time delay without requiring an explicit estimate of each delay. The new algorithm can control unstable and/or non-minimum phase processes; it decouples the loops, both dynamically and at steady state. Consider systems described by \(\ell\)-input/\(\ell\)-output, linear, discrete, randomly disturbed models (1) \(A(z^{-1})Y(t)=\{B^{ij}(z^{-1})z^{- k^{ij}_{\min}}\}U(t)+X(t)\) where U and Y are \(\ell\)-vectors of the deviations from mean system input and output, respectively, \(k^{ij}_{\min}\) is the minimum expected time delay between the i,j output-input pair, and \(\{B^{ij}(z^{-1})(z^{-k^{ij}_{\min}}\}\) represents an \(\ell \times \ell\) matrix polynomial \(B(z^{-1})\). \(A(z^{-1})\) is an \(\ell \times \ell\) diagonal matrix polynomial and \(B(z^{-1})\) is an \(\ell \times \ell\) matrix polynomial. X(t) is an \(\ell\)-vector of disturbances which is assumed to have either the form \(X(t)=C(z^{-1})\xi (t)+d\) or \(\Delta X(t)=C(z^{-1}){\hat \xi}(t)\) where \(\xi\) (t) is an \(\ell\)-vector of sequences of independent \((0,\sigma^ 2)\) variables, d is an \(\ell\)-vector of constant offsets, \({\hat \xi}\)(t) is an \(\ell\)-vector sequence of independent \((m,\sigma^ 2)\) variables, \(\Delta =(1-z^{-1})\), and \(C(z^{-1})\) is an \(\ell \times \ell\) diagonal matrix polynomial. We split \(\{B^{ij}(z^{- 1})z^{-k^{ij}_{\min}}\}\) into two matrix polynomials: \[ \{B^{ij}(z^{-1})z^{-k^{ij}_{\min}}\}=\{B_ D^{ii}(z^{- 1})z^{-k^{ii}_{\min}}\}+\{B^{ij}_{UL}(z^{-1})z^{- k^{ij}_{\min}}\}, \] where \(B_ D(z^{-1})\) is a diagonal matrix polynomial and \(B_{UL}(z^{-1})\) is an upper and lower triangular matrix polynomial. Let \(P(z^{-1})\) and \(R(z^{-1})\) be arbitrary diagonal matrix polynomials and \(Q(z^{-1})\) be an arbitrary \(\ell \times \ell\) matrix polynomial. Using the matrix polynomial identity \(PA^{-1}C=E+\{(FA^{-1})^{ii}z^{-k^{ii}_{\min}}\}\) where E and F are diagonal matrix polynomials, the control law \[ S_ DU(t)=- FY(t)+CRY_ r(t)-\gamma -\{S^{ij}_{UL}z^{(k^{ii}_{\min}- k^{ij}_{\min})}\}U(t) \] is derived, where \(Y_ r\) is a setpoint vector and \(S_ D=EB_ D+CQ_ D\), \(S_{UL}=EB_{UL}+CQ_{UL}\), \(\gamma =E(1)d\). Then, the closed-loop behaviour of the process becomes \[ (B_ DP+Q_ DA)\{y^ i(t+k^{ii}_{\min})\}=B_ DRY_ r(t)+Q_ Dd+B_ DC^{-1}(Ed-\gamma)+ \] \[ +\{[Q_{D}^{ii}B^{ij}_{UL}- B_{D}^{ii}Q^{ij}_{UL}]z^{(k^{ii}_{\min}- k^{ij}_{\min})}\}U(t)+ \] \[ +(B_ DE+CQ_ D)\{\xi^ i(t+k^{ii}_{\min})\}. \] P and \(Q_ D\) are chosen such that the closed-loop poles of the diagonal system are placed at prespecified locations. \(Q_{UL}\) is determined so as to guarantee complete static decoupling, but only to approximate dynamic decoupling.
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references