TITO sistemleri kararlı kılan sabit köşegen kontrolörler

Proses kontrolün ilk ve en önemli amacı sistemi kararlı kılmaktır, performans üzerine yapılacak her eylem bu temel koşul sağlandığı müddetçe bir anlam taşır. Kontrolör tasarımında kullanılan yöntemlerden biri sistemi kararlı kılacak tüm kontrolörleri bulmak ve ardından, tasarımdaki diğer beklentileri sağlamak üzere, bu sınıf içinden uygun bir kontrolörde karar kılmaktır. Bu hedef doğrultusunda yapılanlar şu özellikleri de barındırmalıdır ki, iyi bir kontrolör tasarım sürecinden söz edilebilsin. Kontrolör kırılgan olmamalıdır, yani kontrolör de parametre değişimlerine karşı sistemin kararlılığını ve mümkünse performansını zedelememelidir. Kontrolörün düşük mertebeden olması, parametre ayarlama sürecinde büyük kolaylıklar sağlar, zira ne kadar az parametre o kadar basit bir tasarım süreci demektir. Öte yandan pratik anlamı dolayısıyla da az parametre, ya da düşük mertebe endüstride görev alan uygulamacılar için problemlerin daha hızlı çözülmesi anlamına gelir. Bu çalışmada çok değişkenli kontrol yapıları arasında önemli bir yer tutan karakteristik değerler ve karakteristik değer eğrileri ele alınmış, rasyonel polinomlar cisminde indirgenebilir olan karakteristik denklemler için karakteristik değerlerin reel ekseni kesim noktaları ve bu noktalar civarındaki davranışları üzerinde durulmuştur. Zira bu değerler çok girişli çok çıkışlı sistemleri kararlı kılan kontrolörlerin bulunmasında ya da verilen bir kazanç değeri için kararlılığın analiz edilmesinde hızlı hesaplanabilirlikleri yönünden önem arz etmektedirler. Bu çalışma kapsamında iki girişli iki çıkışlı (TITO) sistemleri kararlı kılan sabit köşegen kontrolörler üzerinde durulmuş, karakteristik değer eğrilerinin reel ekseni kestikleri yerler ve eğrilerin bu noktalardaki geçiş yönlerinden hareket ederek hızlı bir algoritma geliştirilmiş ve bu tür sistemleri kararlı kılan kontrolörler örnek sistem üzerinde sınanarak hesaplanmıştır.

Stabilizing constant diagonal controllers for TITO systems

A control system is an interconnection of components to perform certain tasks and to generate desired output signal, when it is driven by the input signal. In contrast to an open-loop system, a closedloop control system uses sensors to measure the actual output to adjust the input in order to achieve desired output. Most industrial control systems are no longer single-input and single-output (SISO) but multi-input and multi-output (MIMO) systems with a high coupling between the channels. In the design of all dynamical systems stability is the most important property. When a dynamic system is described by its input-output relationship such as a transfer function, the system is stable if it generates bounded outputs for any bounded inputs; bounded-input boundedoutput (BIBO) stability. For a linear, time-invariant system modeled by a transfer function matrix, the BIBO stability is guaranteed if and only if all the poles of the transfer function matrix are in the openleft- half complex plane. A reasonable approach to controller design is to find the set of all stabilizing compensators and then using a member of this set to satisfy further design criteria. A complete parameterization of all stabilizing controllers for a given system was suggested by Youla. An important disadvantage of this parameterization is that the order of the controller cannot be fixed. As a result, the order of the controller tends to be quite high most of the time. Therefore, in the last few years computation of all stabilizing controllers of a given order is examined by several researchers. It is a common fact that it is more difficult to design controllers for MIMO systems because there are usually interactions between different control loops. To overcome this difficulty decentralized controllers are considered which have fewer tuning parameters compared to general multivariable controllers. For example, decentralized PID (P: proportional, I: integral, D: derivative) controllers are widely used in process control due their simplicity and facility in working in case of actuator and/or sensor failure because it is relatively easy to tune manually as only one loop is directly affected by the failure. If a MIMO system described by a n×n transfer-function matrix G(s) is diagonal dominant over the bandwidth of interest, or there exists an input compensator matrix C(s) to achieve diagonal dominance, then the stability and time domain behavior of the system can be inferred from the diagonal elements of G(s)C(s). The relative gain array, the (inverse−) Nyquist array approach, the block Nyquist array method, the Perron-Frobenius scaling procedure and the characteristic locus method are among the analysis and design methods to reduce the interaction in a multivariable system. However, these approaches do not provide the set of all stabilizing controllers. Generalizing the Nyquist stability criterion for MIMO case is particularly important because plotting the characteristic values of the open-loop transfer function enables us to check the stability of the closed-loop system for a gain parameter. A stable characteristic polynomial, becomes unstable if and only if at least one root crosses the imaginary axis. The parameter values of the root crossing form the stability boundaries in the parameter space, which can be classified into three cases: the real root boundary, where a root crosses the imaginary axis at the origin, the infinite root boundary, where a root leaves the left half plane at infinity and the complex root boundary, where a pair of conjugate complex roots crosses the imaginary axes. These stability boundaries separate regions in which the number of closed loop system unstable poles does not change in parameter space. The main objective of this work is to develop an efficient and fast algorithm that can be used in finding all stabilizing constant controllers of diag(k,k)-type for TITO processes. Recall that a TITO system has two characteristic values that are in the field of rational functions, if the characteristic equation is reducible in this field. Hence Nyquist stability criterion can be applied to both of the characteristic values in order to determine the conditions for stability. Recall that the generalized Nyquist theorem requires that the net sum of encirclements of the point -1 by the characteristic values equal to the number of open-loop unstable poles of the system. Hence, it is of special importance to determine where the characteristic locus intersects with the real axis, i.e. where the imaginary part of i( ) λ jw , i = 1, 2 is zero. The direction of these crossings is also important, because the net count of crossings at an intersection point will indicate whether there are closed-loop poles to cross the imaginary axis.

___

  • Ackermann, J. ve Kaesbauer, D., (2003). Stable polyhedra in parameter space, Automatica, 39, 937–943.
  • Barman, J.F. ve Katzenelson, J., (1974). A generalized Nyquist-type stability criterion for multivariable feedback systems, International Journal of Control, 20, 4, 593–622.
  • Gryazina, E.N. ve Polyak, B.T., (2006). Stability regions in the parameter space: D-decomposition revisited, Automatica, 42, 1, 13–26.
  • Ho, M.-T., Datta, A. ve Bhattacharrya, S.P., (1996). A new approach to feedback stabilization, The 35th IEEE Conference on Decision and Control., Kobe, Japan, December 11-13.
  • Ho, M.-T., Datta, A. ve Bhattacharrya, S.P., (1997). A linear programming characterization of all stabilizing PID controllers, IEEE Proc. American Control Conference, Albuquerque, NM, USA, June 4-6.
  • MacFarlane, A.G.J. ve Postlethwaite, I., (1977): Generalized Nyquist stability criterion and multivariable root loci, International Journal of Control, 25, 1, 81–127.
  • Munro, N., Söylemez, M.T. ve Baki, H., (1999). Computation of D-Stabilizing low-order compensators, Control System Centre Report, 882, UMIST, Manchester, UK.
  • Rosenbrock, H.H., (1969). Design of multivariable control systems using the inverse Nyquist array, Proceding of IEEE, 116, 1929–1936.
  • Rosenbrock, H.H., (1970). State-Space and Multivariable Theory, London, Nelson, UK.
  • Shafiei, Z. ve Shenton, A.T., (1994). Tuning of PIDtype controllers for stable and unstable systems with time delay, Automatica, 30, 1609–1615.
  • Shafiei, Z. ve Shenton, A.T., (1997). Frequencydomain design of PID controllers for stable and unstable systems with time delay, Automatica, 33, 12, 2223–2232.
  • Skogestad, S. ve Morari, M., (1989). Robust performance of decentralized control systems by independent designs, Automatica, 25, 1, 119–125.
  • Söylemez, M.T., Munro, N. ve Baki, H., (2003). Fast calculation of stabilizing PID controllers, Automatica, 39, 121–126.
  • Söylemez, M.T. ve Üstoğlu, İ., (2006). Designing control systems using exact and symbolic manipulations of formulae, International Journal of Control, 79, 11, 1418–1430.
  • Söylemez, M.T. ve Üstoğlu, İ., (2007). Polynomial control systems, IEEE Control System Magazine, 27, 4, 124–137.
  • Youla, D.C., Bongiorno, J.J. ve Lu, C.N., (1974). Single-loop feedback stabilization of linear multivariable plants, Automatica, 10, 2, 159–173.
  • Youla, D.C., Jabr, H.A. ve Bongiorno, J.J., (1976). Modern Wiener-Hopf design of optimal controllers – Part II: multivariable case, IEEE Transactions on Automatic Control Theory, 21, 3, 319– 338.