Geri Dön

Asenkron makina kontrolü için yapay sinir ağı tabanlı rotor akışı gözlemcisi

Başlık çevirisi mevcut değil.

  1. Tez No: 75297
  2. Yazar: ASLI AYLA ÇAKIRGÖZ
  3. Danışmanlar: PROF. DR. EMİN TACER
  4. Tez Türü: Yüksek Lisans
  5. Konular: Elektrik ve Elektronik Mühendisliği, Electrical and Electronics Engineering
  6. Anahtar Kelimeler: Belirtilmemiş.
  7. Yıl: 1998
  8. Dil: Türkçe
  9. Üniversite: İstanbul Teknik Üniversitesi
  10. Enstitü: Fen Bilimleri Enstitüsü
  11. Ana Bilim Dalı: Elektrik Mühendisliği Ana Bilim Dalı
  12. Bilim Dalı: Belirtilmemiş.
  13. Sayfa Sayısı: 121

Özet

Tezde asenkron makinanın vektör denetimine ilişkin rotor akısı gözlemcisinin yapay sinir ağı ile gerçeklenmesi ele alınmıştır. Rotor akısı gözlemcisinin kullanılmasındaki amaç, seçilen asenkron makina vektör denetiminde, kontrolün gerçeklenmesi için rotor akısı uzay fazörü büyüklüğünün ve açısının bilinmesi gereğindendir. Ayrıca, kontrol büyüklüklerinin herhangi bir algılayıcı kullanılmadan elde edilmesi yoluna gidilmiştir. Tez de asenkron makinanın rotor alan yönlendirmeli denetimine ilişkin detaylı açıklamalar yapılmış, asenkron makinaya ve vektör denetimine ilişkin simulasyon sonuçlan verilmiştir. Ayrıca yapay sinir ağı tabanlı bir gözlemci oluşturulmuş ve bu gözlemci ile herhangi bir algılayıcı kullanmadan rotor akısı yönlendirmeli vektör denetimi için gerekli olan büyüklükler rotor akısı genliği ve açısı elde edilmiştir.

Özet (Çeviri)

In this work, a neural rotor flux observer for induction motor control is examined and various simulation are given. The asynchronous motor, also known induction motor is the most commonly used in drive system. The squirrel cage rotor induction motor has a cage winding in the rotor. The bars and shorting rings are made of copper, brass or aluminium and, as a rule, they are not insulated from the core. This makes it possible to fabricate windings by casting, which makes for very simple, inexpensive and reliable motor design. The control and feedback signal processing of drives is considerably more complex than the traditional dc drives, and this complexity is compounded if higher performance is demanded. One reason for the complexity dynamics (d-q) model can be described by a high-order nonlinear multivariable state-space equation [1]. An induction motor can be described by the following state equations in the stationary reference frame [6]: dt Vr An A12 A2ı A22 Us =Ax + BUs i9 = CX Where i3 = [isd isq]T : stator current \j/r = [v|/"i \j/«JT :rotorflux Us= [Usd Usq]T : stator voltage Rs (1-a) An = -i + } 1 = 3,11 1 crLs aTr XIIlM 1 A12 = {( ) I - Wr j} = ari2l + &i\2 J CLsLr Tr M A21 = I = 8x21 I Tr 1 A22 = 1 + Wr J = ^22 I îte J Tr B, I =bıl oLs C= 1 0 I = 1 0 0 1 J = 0 -1 1 o Ra, Rr : Stator and rotor resistance Lr, Ls : Stator and rotor self inductance M : Mutual inductance M2 o : Leakage coefficient, a = 1 Ligler Tr : Rotor time constant, Tr = - Rr wr : Motor angular velocity.The vector of field-oriented control tecnique brought on a renaissance in modern high- performance control of A.C drives. This control method has found wide acceptance in applications such as paper mills, textile mills, steel rolling mills, machine tools, servos and robotics. With vector or decoupling control, the dynamics of A.C drives is similar to that of D.C drives and with currrent control, the conventionel stability limit of A.C machine does not arise. The aim of vector control methods is to control an induction machine in both transient and steady state conditions by using mathematical model in either synchronously or any orbitrary speed rotating reference frame. By applying vector control methods, it is possible to reach a higher performance and phase the control quantities must be thought in vector control methods. Field orientation İs based on the knowledge of the rotor flux vector. The induction machine models can be given with the help of space phasor theory is expressed as ; Uffl u, ay uc u, jy isy *n Where stator and rotor quantities are given in quadrature axes. Using the %&, i»,, \|/ri, \j/n, and wr as state variables induction machine model is simulated using Matlab- Simulink Programme. In field oriented drives, the sending rotor flux as a vector ( both modulus and angle ) is realised two different tecniques. In the first method, the direct or feedback method, which was developed by Blaschke, depends on unit vector generation from the machine terminal voltages. The flux vector is directly sensed by using Hall probes or test coils mounted on the stator winding of machines. In the second method, the indirect or feedforward method, which was developed by Hasse ¦-, ıs based --on- calculating the rotor flux vector from dynamics machine equations for field orientation constraints. But in the second method the controller is highly dependent on machine parameters. An original tecnique for estimating the magnetic rotor flux of an induction motor drive based on feedforward neural networks is examined in this study. The term neural network is analogous to the nervous system in the human brain, where a large number of nerve cells are interconnected by input dentrites and out put axons. The input parallel signals from a layer of cells in parallel though the axons.A neurocomputing network can be looked on as a patern processing system, where the input signal patern is processed to a desired output pattern. A neural net has the capability to learn that consist of varying of weighting coefficients for the input signals of a layer for the desired transfer characteristics. A simple example of an analog neural network is a multiinput opamp summer with varying input resistors. A neuron can be represented in a shematic way as an element with several inputs and a single output, as shown in fig. 1. The eloboration of the single neuron unit consists in a weighted sum of input variables and in a nonlinear transformation by means of the activation function. Synaptic connections Nearoa's processing node Multiplifi^re Weights Figure 1. Neuron scheme as a mathematical element The coefficients Wiq are referred as synaptic weights ; they determine the effect of the i-th input that is connected to the sum node of the q-th neuron. A A li*2 ^k+i lk*l Lk*l + a? 4 - C* J] %k*1 w. '_k*1 K, *»2 K. iJk*Z 1 Figure.2. Estimator RepresentationFigure 2 shows the estimator and cost computation scheme, in cascade configuration. Owing to the need of a simple graphical representation, the input Uk and the speed wt are not displayed.; moreover, in such sheme, cp and i replace and is respectively. In addition, under the hypothesis to consider the instant in which the information set Ii is avaiable, the second subscript i is omitted from all the symbols. [3] The backpropagation algorithm is the basic algorithm for learning in multilayer perceptron. In this algorithm, an error function is defined and equal to the mean square difference between the desired output and the actual output of the network. In order to minimize this error function, the backpropagation algorithm uses agradient descent tecnique. The network is trained by initially selecting learning rate and small random weights and biases and then presenting all training data repeatedly. Weights and theresholds are adjustment continues until weights and biases converge and error function is reduced to an acceptable value. The purpose of momentum method is to accelarate the convergence of the backpropagation algorithm. The method involves supplementing the current weight adjustments with afraction of most recent weight adjustments with a fraction of most recent weight adjustment. Without momentum a network may get stuck in a shallow local minima. With momentum a network can slide through such a minimum. Sytem identification, one of sytem fündemental issues in system theory, is concerned with the mathematical representation of the behavior of a system. More specifially, system identification seeks to derive the mathematical expression which relate some sets of observed input and output data of the prosess. Identification may be carried out off-line, with a continous monitoring of input-output data during operation. System identification includes four step: 1. Input / output data acquisition under an experimentation protocol, 2. Choice of model structure, 3. Estimation of the model parameter, 4. Validation of identified model (structure and values of parameters). It appears that in the future, the elements of expert systems, fuzzy logic and neural networks will be combined to gain performance optimization of power electronics systems.

Benzer Tezler

  1. Tez No

    65129

    Neural Network based flux observer for a field oriented controlled induction machine

    Rotor akısı yönlendirmeli denetimli bir asenkron makina için yapay sinir ağı tabaklı akı gözlemcisi

    V.M.İSTEMİHAN GENÇ

    Yüksek Lisans

    İngilizce

    İngilizce

    1997

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve KontrolBoğaziçi Üniversitesi

    PROF. DR. OKYAY KAYNAK

  2. Tez No

    675955

    Modelling longitudinal motion of an electric vehicle and wheel slip control through NN based uncertainty prediction

    Elektrikli aracın boyuna hareketinin modellenmesi ve yapay sinir ağı tabanlı belirsizlik kestirimli tekerlek kayma kontrolü

    DUYGU ÖZYILDIRIM

    Yüksek Lisans

    İngilizce

    İngilizce

    2021

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrolİstanbul Teknik Üniversitesi

    Mekatronik Mühendisliği Ana Bilim Dalı

    PROF. DR. OVSANNA SETA ESTRADA

  3. Tez No

    882559

    Asenkron motor eşdeğer devre parametrelerine etki eden faktörlerin makine öğrenme yöntemleri ile belirlenmesi

    Determination of factors affecting induction motor equivalent circuit parameters by machine learning methods

    ABDULLAH CEM AĞAÇAYAK

    Doktora

    Türkçe

    Türkçe

    2024

    Elektrik ve Elektronik MühendisliğiKonya Teknik Üniversitesi

    Elektrik ve Elektronik Mühendisliği Ana Bilim Dalı

    DR. ÖĞR. ÜYESİ HAKAN TERZİOĞLU

  4. Tez No

    39286

    Yapay sinir ağlarında öğrenme algoritmalarının analizi

    Analysis of learning algorithms in neural networks

    SEVİNÇ BAKLAVACI

    Yüksek Lisans

    Türkçe

    Türkçe

    1994

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrolİstanbul Teknik Üniversitesi

    DOÇ.DR. LEYLA GÖREN

  5. Tez No

    169917

    Rüzgâr enerjisi dönüşüm sistemlerinde kullanılan generatörlerin denetimi

    Controlling of generators in wind energy conversion systems

    ZAFER ÖZER

    Yüksek Lisans

    Türkçe

    Türkçe

    2005

    Elektrik ve Elektronik MühendisliğiKahramanmaraş Sütçü İmam Üniversitesi

    Elektrik-Elektronik Mühendisliği Ana Bilim Dalı

    Y.DOÇ.DR. AHMET SERDAR YILMAZ

Geri Dön