Geri Dön

Endüstriyel manipülatörlerin nonlineer kontrolu

Nonlinear control of industrial manipulators

  1. Tez No: 21980
  2. Yazar: AHMET ZORLU
  3. Danışmanlar: DOÇ. DR. CAN ÖZSOY
  4. Tez Türü: Yüksek Lisans
  5. Konular: Makine Mühendisliği, Mechanical Engineering
  6. Anahtar Kelimeler: Belirtilmemiş.
  7. Yıl: 1992
  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: 136

Özet

ÖZET: Son yıllarda bilgisayarların niteliklerinde meydana gelen olumlu gelişmelerin, nonlineer sistemlerin kontrolunda nonlineer kontrol algoritmalarının kullanımını mümkün kılması, lineer kontrol algoritmaları ile kontrolda meydana gelen bazı yetersizlikleri ortadan kaldırma imkanı vermektedir.. Nonlineer bir sistem olan robot manipülatörlerin kontrolunda da bu durum geçerlidir. PD kontrolü gibi klasik bir lineer kontrol düşük hızlı yörüngelerde kafi gelmesine rağmen yörünge hızı arttıkça izleme hataları da artmaktadır. Bunun yerine sisteme uygun bir nonlineer kontrol uygulandığında, bu sakıncalar giderilebilmektedir. Nonlineer kontrolda, gerçek sistem ile kontrol sinyalinin üretiminde kabul edilen sistem arasında farklılıklar varsa, diğer bir deyişle modelleme hataları ve bilinmeyen bozucular mevcut ise nonlineer kontrolün da performansının kötüleseceği açıktır. Bu kötüleşmeyi kısmen de olsa gidermek için bir robust kontrol algoritmasına ihtiyaç duyulabilir. Bu çalışmada PD kontrolü, nonlineer kontrol ESİ ve nonlineer robust kontrol E 181 algoritmaları incelenmiş, sonuçta nonlineer kontrolün PD kontroluna, nonlineer robust kontrolün da modelleme hataları ve bilinmeyen bozucular altında nonlineer kontrola üstünlüğü ortaya çıkmıştır.

Özet (Çeviri)

NONLINEAR CONTROL OF INDUSTRIAL MANIPULATORS SUMMARY Industrial use of robot manipulators includes material handling» assembly» spray painting» welding etc. It is desired that for employing robot manipulators in industry» they have to do their tasks cheaper, more accurate and less time consuming according to the human worker. To achieve this goal, robot hand must have ability for accurately tracking of trajectory and positioning. The dynamic equation that can be derived from Lagrange- Euler formulation of robot manipulators which is a nonlinear system contains inertial forces, coriolis and centrifugal forces and torques that are applied to joints. coriolis and centrifugal forces and inertia related matrix that is not a diagonal matrix and used for computing inertial forces can lead to important coupling effects. Nowadays, PD, PI, PID control methods that have been widely used in many places have been applied to robot manipulators. The control signal of each joint, which is computed by using these control methods is produced from its position and velocity values that are measured with sensing devices and reference values. In other words» the effects of other joints and links are not considered. Performance of these control methods that is sufficient at low speed robot control deteriorates at rapid motion that coupling effects become prominent. PD, PI, PID control methods are not suitable for rapid motion of robot manipulators. vmSince coupling effects is formed by gravity, coriolis and centrifugal forces terms of dynamical equation of the robot manipulator, it is reasonable to apply feedforward control to cancel the coupling effects and then apply feedback control to account for external disturbance. This algorithm is known as computed torque method. However, when the parameters in the dynamical equation of robot manipulator that is used for producing the control signal differs from the actual values of the system, satisfactory performance can not be achieved. In this study, robot dynamics in Lagrange- Euler formulation, general nonlinear concept, application of the general nonlinear concept to the robot manipulators [S3 and nonlinear robust control C183 are examined. Nonlinear control signal results in linearisation and decoupling of system and assigning of closed- loop poles of joints easily. Nonlinear control cancels coupling effects. Nonlinear robust control increases the resistance of the system against the modelling errors and unknown disturbances. Nonlinear robust control consist of two parts: First, nonlinear control part which decouples and linearizes the closed- loop system and enables us to assign closed- loop poles of each decoupled joint system. Second, linear control part which enables us to suppress effects of uncertainties (modelling errors or unknown disturbances. ) For the simulation, a program is developed in Fortran 77 programming language. Data which are required for simulation program contain Denavit-Hartenberg parameters of the robot manipulator, the masses of the links and the mass of the pay load, the initial conditions of the joint positions and velocities, the initial time of the simulation, the integration step number of the simulation, writing frequency of the outputs, gravitational vector, position vectors of the center of the link masses according to the joint coordinate systems, squares of the gyration radii of the links, viscous friction constants, Coulomb friction constants, choosing of the control algorithm and trajectory, constants of the trajectories, numerical values of the poles for the nonlinear control part and the linear control part which are required for the nonlinear control and the nonlinear robust control, constants of the position gains and velocity gains of PD control, assumed xxtorque limits that motors can deliver etc. The data are read from a data file at the beginning of the simulation. Lagrange-Euler method is used for modelling the system. Dynamics of links and pay load are seperately considered. Payload is adopted that it is a point mass and firmly fixed to the hand of robot manipulator. The payload dynamics is computed through forward kinematics and their first and second derivatives and Jacobian matrix. Then it is added to dynamics of the links that is computed by using Lagrange-Euler formulation. In the simulation, Runge-Kutta numerical integration method is used. And integration incremental step size is adopted as 0.001 second. Simulation time is chosen as eight seconds. It Is accepted that the system has viscous and Coulomb friction. Dynamics of motors are not considered. A three link manipulator is chosen for the simulation. The manipulator has three revolute joints. Three trajectories that have cubic, exponential and cosine form are used for comparing the performance of PD control with that of the nonlinear control. Simulation results are drawn graphs that shows positions, velocities, tracking errors and torques of joints versus time. Cosine trajectory is used for comparing nonlinear control with nonlinear robust control. And in the simulation, when the time reachs 3.14 seconds, the link masses are increased two times and the payload is increased three times. So that modelling errors and unknown disturbances occurred. Then the system behaviour is examined. From all of these simulations, following conclusions are achieved: When the nonlinear control signal that is achieved from the nonlinear control theory is applied to the robot manipulator that is a nonlinear system, the system is linearized and decoupled. For each joint, two poles can be assigned arbitrarily. The nonlinear control does not need velocity and acceleration references, only position reference trajectory is sufficient. When closed- loop poles of the each joint are slided toward the lefthand side of the complex plane, the sum of the square of position tracking errors are decreased. By choosing appropriate poles, damping ratio of the closed- loop system can be assigned.When. PD control is compared with the nonlinear control, the nonlinear control has less tracking error according to PD control. In other words, nonlinear control enables the manipulator to follow the trajectory more accurately. In PD control method, actual outputs oscillates around the reference trajectory. The final errors in the PD control simulations is larger than that of the nonlinear control for cubic and exponential trajectories. Yet, in the PD control simulations, instantaneous and important changes are seen on velocity graphs. But in the nonlinear control, there is not such a kind changes in them. And velocities regularly increase and decrease. In PD control, torques that is applied to joint actuators also have rapid descents and ascents. So this descents and ascents cause control chattering. Such a control signal also cause rapid descents and ascents on velocity graphs. In the nonlinear control simulations, such a kind changes are not seen. Since control signal has not rapid descents and ascents, the nonlinear control requires less torque value to follow the trajectory according to PD control. This means that if the nonlinear control is applied, smaller motor which is lighter and cheaper can be chosen. Smaller motor also enables the manipulator to carry much more pay load. Rapid descents and ascents on the velocity graphs indicates rapid descent and ascent on the acceleration graphs. This type accelerations cause frequently changing forces and this forces bring about fatique of the manipulator material. If the frequency of these changes in forces is close to the natural frequency of the manipulator, there will be resonance which can cause manipulator damage. XIThe nonlinear control has a good performance without modelling errors and unknown disturbances. When the modelling errors and unknown disturbances are introduced, the performance of the nonlinear control deteriorates. To remove the deterioration, the nonlinear control [53 is combined with the multivariable robust servomechanism theory E 163 and new control theory is introduced which can be called nonlinear robust control C183. The nonlinear robust control has less tracking errors under the modelling errors and unknown disturbances according to the nonlinear control. Especially.,, when the manipulator movces at high speed-, the velocity graphs of the nonlinear control have instantaneous changes and control signals also have considerably changes according to the nonlinear robust control. It is easy to compute the input signals of the manipulator by using PD control, but it is difficult to adjust PD control constants that are position gain Kp and velocity gain Kv. In addition, coupling effects that reach important values at high speed motion deteriorate the performance of PD control. Even they can make PD control unstable. Since input signals of the nonlinear control and the nonlinear robust control are based on the system model, there is no problem at high speeds. As computing of the input signals of both the nonlinear control and the nonlinear robust control is inefficient in terms of computing time by using Lagrange-Euler formulation, Newton-Euler recursive formulas are suitable in terms of computing time in practice. But, even Newton-Euler recursive formulas are used, they may require faster microprocessors according to the microprocessors that can be used for PD control. Eventually, the nonlinear control and the nonlinear robust control that are investigated this study are suitable methods for controlling robot manipulator that is a nonlinear system. XX i

Benzer Tezler

  1. Tez No

    421264

    Staubli RX160 manipülatörün sensörsüz kuvvet kontrolü

    Sensorless force control of Staubli RX160 manipulator

    ONUR TAŞKIRAN

    Yüksek Lisans

    Türkçe

    Türkçe

    2015

    Mekatronik Mühendisliğiİstanbul Teknik Üniversitesi

    Mekatronik Mühendisliği Ana Bilim Dalı

    DOÇ. DR. ZEKİ YAĞIZ BAYRAKTAROĞLU

  2. Tez No

    100704

    Yapay sinir ağları ile robotlarda hareket kontrolü

    Motion control of robots with artificial neural networks

    HAKAN ARSLAN

    Doktora

    Türkçe

    Türkçe

    1999

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrolİstanbul Teknik Üniversitesi

    PROF.DR. AHMET KUZUCU

  3. Tez No

    634597

    Trajectory generation for industrial robots in presence of wrist singularity

    Endüstriyel robotlarda bilek tekilliği ve yörünge planlaması

    CANSIN AKARSU

    Yüksek Lisans

    İngilizce

    İngilizce

    2020

    Makine Mühendisliğiİstanbul Teknik Üniversitesi

    Mekatronik Mühendisliği Ana Bilim Dalı

    DOÇ. DR. ZEKİ YAĞIZ BAYRAKTAROĞLU

  4. Tez No

    823534

    6 serbestlik dereceli endüstriyel robotlar için sürü zekâlı algoritmalara dayalı optimal yörünge planlaması ve kontrolü

    Optimal trajectory planning and control based on swarm intelligence algorithms for 6 dof industrial robots

    HASAN KARCI

    Doktora

    Türkçe

    Türkçe

    2023

    Elektrik ve Elektronik MühendisliğiKocaeli Üniversitesi

    Elektronik ve Haberleşme Mühendisliği Ana Bilim Dalı

    PROF. DR. ALİ TANGEL

  5. Tez No

    579513

    Design and control of a winch driven grasping mechanism for a quadrotor unmanned aerial vehicle

    Dört rotorlu insansız hava aracı için makaralı yük alma-bırakma mekanizması tasarımı ve kontrolü

    MEHMET OKAN GÜNEY

    Yüksek Lisans

    İngilizce

    İngilizce

    2019

    Makine Mühendisliğiİstanbul Teknik Üniversitesi

    Makine Mühendisliği Ana Bilim Dalı

    DOÇ. DR. ERDİNÇ ALTUĞ

Geri Dön