Geri Dön

Adaptif kontrol sistemleri ve bir mikrokontrolör ile simülasyonu

Adaptive control systems and simulation by a microcontroller

  1. Tez No: 14423
  2. Yazar: CANAN MÖRÜ
  3. Danışmanlar: DOÇ.DR. LEYLA GÖREN
  4. Tez Türü: Yüksek Lisans
  5. Konular: Elektrik ve Elektronik Mühendisliği, Electrical and Electronics Engineering
  6. Anahtar Kelimeler: Belirtilmemiş.
  7. Yıl: 1991
  8. Dil: Türkçe
  9. Üniversite: İstanbul Teknik Üniversitesi
  10. Enstitü: Fen Bilimleri Enstitüsü
  11. Ana Bilim Dalı: Belirtilmemiş.
  12. Bilim Dalı: Belirtilmemiş.
  13. Sayfa Sayısı: 98

Özet

ÖZET Bu çalışmada adaptif kontrolör ve sınıflandırılmala rı genel olarak incelendikten sonra adaptif kontrolörle rin bir özel sınıfı olan model referans adaptif kontro lörler CMRAO incelenmiştir. Model referans adaptif sis temlerin üç sınıfının sistem kontrol uygulamaları açısın dan yorumları açıklanmıştır. Bahsedilen üç sınıf, x Belirleme hatası yöntemi * Giriş hatası yöntemi * Çıkış hatası yöntemi dir. Tasarımlarda, kutup kaydırma ve sıfır silme karakte ristiklerini incelemek için klasik kontrol tekniklerinden yararlanılmıştır. Tek-giriş tek -çıkışlı lineer sistemler için mikrobilgisayar MRAC *ın üç ana bloğu olan sistem, referans model ve kontrolör, bilgisayarla simüle edilip belirleme hatası yöntemine göre kontrol işareti gerçek zamanda Con-line) parametre belirlemesi yapılarak hesap lanmıştır. Bu amaçla C programlama dilinde bir program yazılmıştır. Program, ayrıca HPC 18083 mikrokontrolörüne uyarlanıp kullanılabilir. Parametre belirleme için en küçük kareler yöntemi C leas t squares method kullanılmış tır. Uygulamalardaki sistem kazancı, parametreler dek i de ğişimler gibi pratik sınırlamaların giriş, çıkış ve belir leme hata yöntemlerinin performansını ne ölçüde etkiledi ği incelenmiştir. Sonuç olarak, pratik ve güvenilir bir mikrobilgisayar MRAC *ın, seri-paralel referans model içeren geliştirilmiş bir belirleme hatası tasarımı ve en küçük kareler yöntemi ni kullanan bir ayarlama mekanizması ile oluşturulabilece ği görüşü üzerinde durulmuştur. Cvi>

Özet (Çeviri)

ADAPTIVE CONTROL SYSTEMS AND SIMULATION BY A MICROCONTROLLER SUMMARY Adaptive control is an important area of modern control dealing as it does with the control of the systems in the presence of uncertainties, structural perturbations and environmental variations. There has been a recent upsurge of interest in this field, since adaptive controllers can be implemented simply because of the availability of increasingly versatile digital hardware. Adaptive control techniques have also benefited from the steady and even spectacular reduction in the cost/performance ratio of microelectronic devices in recent years. This has resulted in a wide variety of industrial applications in situations which were not considered easily implementable earlier. Furthermore, many theoretical problems that had baffled researchers have also been solved during the past few years. There have been major advances in the design of adaptive control systems, especially in terms of global stability. Several adaptive control strategies have been successfully applied in diverse practical problems. In this thesis, the adaptive control and why and when adaptive control is used have been introduced. Some important landmarks and early work are discussed. Although there is no standardized definition for adaptive control, several purposed definitions of various terms occurring in this field have been given to give an idea of the scope and objectives of adaptive control. After discussing the essential aspects of adaptive control, there have been classification of these adaptive control systems, in particular, as model reference adaptive control systems. (vii)In recent years model reference adaptive control (MRAC) has became a very efficient systematic method for controlling plants with unknown ( or partially known ) parameters- In this procedure the objectives of control are specified in terms of a reference model and design problem involves the determination, from all available on-line data, of the value of either the adjustable parameters or the control input such that the error between the plant and model outputs approaches zero asymptotically. Different configurations of MRAC systems has been introduced in this study. Technological advances in large-scale (LSI) integration and integrated circuit (IC) design and fabrication are increasing the capabilities of microprocessors and pertient interface while reducing their cost. These advances offer the control engineer both low-cost computing power and the prospect of implementing control theoretic methods which have been developed over the past thirty years. This study focuses on the classical control interpretations of three classes of model reference adaptive controllers (MRAC) for process control apl icat ions. These three classes are * Identification error method * Output error method * Input error method These MRAC are compared in a common framework from the viewpoint of microcomputer implementation. Techniques of classical control engineering are applied to examine the adaptive pole shifting and zero cancellation characteristics of each design. It is demonstrated that the choice of either a full-parallel (FP) or a series- parallel (SP) reference model in the adaptive controller dramatically affects the performance of the closed-loop MRAC. The distinguishing features of gradient and least squares adjustment mechanisms can be seen in the context of practical considerations imposed by the microcomputer. Advanced technological processes, such as high-performance aircraft and robots, must be controlled accurately over the wide range of operating conditions. This environment establishes the characteristics of the feedback system which are required to control the process whit in the design specifications. Design of such controllers is a formidable engineering problem and fixed controllers may compromise the performance of the process (viii)over the entire operating range. Adaptive controllers can alternative these design problems. Since adaptive controllers are self-tuning, the performance of the process remains at the specified level in spite of changes in, or lack of knowledge of, the process characteristics. Through advances in large- scale integration technology, microcomputers are inexpensive and powerful tools for implementing adaptive controllers. Adaptive control algorithms that are effective in the face of practical constraints remain to be developed. While designs of such microcomputer adaptive controllers emerge from a spectrum of approaches, relationships between these designs are lacking in the literature. Little attention is paid to practical constraints that arise in microcomputer implementation. The practical constraints may be imposed by the process, microcomputer, sensors or actuators. Process characteristics dictate the speed of the adjustment » mechanism and the quality of the control signal. A fast process requires an adjustment mechanism whit similar dynamic characteristics. Processes with high order dynamics, such as robot arms, can not tolerate noisy control signals. The microcomputer limits the magnitude of the control signal and sets the minimum cycle time for the algorithm. Thus, algorithm complexity is a key consideration. Errors in the available signals can degrade the performance of algorithms that rely on accurate information. Feedback control involves determining someting about the plant state from plant outputs and using this information to generate control inputs that force the plant to behave in a desired way. It is the use of plant outputs that forms a feedback system. For a plant with known model, the plant state summarizes everything there is presently to know about the plant, so the most information that can be derived from the outputs is the state of the plant. If it is possible to get information about the plant from its inputs and outputs, this information can be used to generate the suitable contol. In this thesis, a method for determining the parameters of a completely observable plant from its outputs and inputs in a finite number of steps is used. The method that is used for this purpose is the least square method. (ix)Measurements of the input and the output of a system can be used to determine the coefficients of the system's input-output equations or equvalently the coefficients of the system transfer functions. This is called system identification. It can be applied to all or part of an existing plant to aid in developing a plant model before controller design is done. Or it can be done as part of the control strategy in what is called adaptive contol. The second one is aimed in this thesis. In adaptive control, the plant is repetetively identified and the controller is changed if the measurements indicate that the plant has chanced. The measurements and appliying the desired control signal to the plant is done by the analog to digital and digital to analog converters in a microcomputer control strategy. The input-output relations of a system are unchanged by any nonsingular transformation of state variables, so there are many state variable models that could account < for the same data. Only the transfer functions or some equvalent set of unique parameters can be found from the system inputs and outputs. In parameter identification, we may have more equations than the minimum number necessary to uniquely characterize a system. Slight errors in measurements generally result in an o ver determined set of equations that is (slightly) inconsistent. For a set of linear algebraic equations with n unknowns, one solution method is to simply solve the first n linearly independent and consistent equations and to ignore any additional equations. A better method is to make use of all the available equations and find an estimate x that is a“best fit ”to all the equations. For this reason least squares algorithms are applied to such problems. The basic least squares problem involves an överde ter mined set of linear algebraic equations H x = z where the matrix H is m x n and m > n. The equations are inconsistent and so have no solution, so we seek an aproximate solution vector x such that the sum of the squares of the errors between the actual known z and the knowns Hx necessary for the equations to be consistent, A A ^ A O O J(x) = ( z-Hx ) ( z-Hx ) =v. +v“+... (x)is minimized, that in That is an estimate x is to be found such ÖJ â x ) = âj dx. âj n H - f E b. q (İ 1} 1 u{j) n E *-i=2 n f ”A -(i-1)“I, ~ L E ai q J y( 3) (XX)u(j) = CHb±, b±, Q) Tt b”n A (J) - f E b q (1 1} 1 u(j) 4=2 x J - f^.q-*1-1' ] y(J, 4=1 J where Ci(.b±, b±, Q) b* + Q2 where b and a coefficients are nonlinear functions of process parameters. q is the backward-shift or delay operator. b is the estimate value of the process coefficients at the (j-l)-th sampling instant. x(j) is the reference model output. When the b = b and x (j) = x (j+1), then the control signal for a second order process has became to a form as follows u(j) bl + Q x(j+l) -b2u(j-l) - a^yCj) - a2y{j-l) It is usually desired that the process follows the reference model output with two cycle delay. In this case x(j) = r{j-2) where r(j) is the reference signal. Then x{j+l) = r(j-l) The resulting control signal equation that is used in the programme is u(j) = bl +Q r(j-l) -b2u(j-l) - a^yU) - a2y(j-l) (xii)the process which is derived by this control signal is modeled as follows y(j> = ajyU-1) + a2y{j-2) + b^j-l) + t>2u(j-2) These equations are used in the programme. When the difference between process output and identifier output excess a limit value, programme calculates the new parameters of the process and finds a new control signal value with new parameters and then apply this signal to the process. For practical applications these MRAC are subject to drifting process gains and parameters, a limited dynamic range for the control signal transmitted by the D/A, and inaccurate measurements and computations induced by the finite word length of the microcomputer. These practical constraints deteriorate the performance of the input and output error methods. when the control signal saturates in the D/A, the input and output error methods can cause erroneous action by the adjustment mechanism. The identification error method is immune from these problems. Since the microcomputer accentuates measurement and computational errors, the FP reference model is impractical for controlling low-pass processes. The high-pass characteristic introduced by the FP reference model amplify high frequency disturbances, whereas the SP reference model does not. The SP reference model increases the stability margin of the closed loop MRAC over that of the FP reference model. Measurement errors also influence the adjustment mechanism. The high gains required for gradient mechanisms to maintain reasonable tracking accuracy are sensitive to noise and can destroy the quality of the control signal. Least squares mechanisms filter measurement noise over successive samples. On the basis of this review, a practical and robust microcomputer MRAC would consist of an enhanced identification error design incorporating a SP reference model and least squares adjustment mechanism. (xiii)

Benzer Tezler

  1. Tez No

    363575

    Comparison of three IPMSM sensorless position estimation methods through simulations and experiments

    İç sabit mıknatıslı senkron motorlarda üç sensörsüz konum tahmini yönteminin simülasyonlar ve deneyler aracılığı ile karşılaştırılması

    KAAN BAYKA

    Yüksek Lisans

    İngilizce

    İngilizce

    2014

    Elektrik ve Elektronik Mühendisliğiİstanbul Teknik Üniversitesi

    Kontrol ve Otomasyon Mühendisliği Ana Bilim Dalı

    YRD. DOÇ. DR. ALİ FUAT ERGENÇ

  2. Tez No

    725981

    Modular kinetic system proposal with responsive design approach for acoustic paneling systems

    Tepkisel tasarım yaklaşımı ile modüler kinetik akustik panel sistemi önerisi

    SUAT BATUHAN ESİRGER

    Yüksek Lisans

    İngilizce

    İngilizce

    2022

    Mimarlıkİstanbul Teknik Üniversitesi

    Bilişim Ana Bilim Dalı

    PROF. DR. MİNE ÖZKAR KABAKÇIOĞLU

  3. Tez No

    78133

    Klasik kontrol yöntemleri ve bulanık denetimin bir PMDC motorda denenerek sonuçların karşılaştırılması

    Experimental evaluation of the classical and fuzzy control method on PMDC motor

    CELAL SAMYELİ

    Yüksek Lisans

    Türkçe

    Türkçe

    1998

    Makine MühendisliğiKaradeniz Teknik Üniversitesi

    Makine Mühendisliği Ana Bilim Dalı

    DOÇ. DR. LEVENT GÜMÜŞEL

  4. Tez No

    421267

    An effective adaptive and fault tolerant controller for helicopters

    Helikopterler için etkin adaptif ve arıza toleranslı kontrolör

    MUHAMMED ALİ NUR ÖZ

    Yüksek Lisans

    İngilizce

    İngilizce

    2016

    Elektrik ve Elektronik Mühendisliğiİstanbul Teknik Üniversitesi

    Kontrol ve Otomasyon Mühendisliği Ana Bilim Dalı

    PROF. DR. FİKRET ÇALIŞKAN

  5. Tez No

    878656

    Guidance, navigation and control of an autonomous underwater vehicle

    Otonom bir su altı aracının güdüm, seyrüsefer ve kontrolü

    MEHMET AVİNÇ

    Yüksek Lisans

    İngilizce

    İngilizce

    2024

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrolİstanbul Teknik Üniversitesi

    Uçak ve Uzay Mühendisliği Ana Bilim Dalı

    PROF. DR. CENGİZ HACIZADE

Geri Dön