LEE- Sistem Dinamiği ve Kontrol-Yüksek Lisans
Bu koleksiyon için kalıcı URI
Gözat
Son Başvurular
1 - 3 / 3
-
ÖgeOn control of direct-drive permanent magnet synchronous machines in micromobility applications(Graduate School, 2023-07-12)In micromobility applications with direct drive powertrain configuration, the traction torque and speed are directly generated by a Surface-mount Permanent Magnet Synchronous Machine (SPMSM), and no other transmission components are utilized, thus the operating region in which the motor must reliably operate is excessively wide. Hence, maintaining satisfactory motor control performance in that wide torque-speed envelope of an SPMSM is a challenge due to the increasing nonlinearities outside the nominal operating region. This thesis deals with the design and analysis of reliable and high-performance motor control systems for direct-driven micromobility SPMSMs. Throughout this thesis, a motor control system is designed for direct-driven SPMSMs to establish high-performance and reliable motor control operation on micromobility platforms. To achieve robust and satisfactory speed control of SPMSM by maintaining reliability, the design and tuning parameters selection of voltage feedback field weakening vector control structure with Cascaded PI (CPI) is explained. A special emphasis is put on the implementation issues and environmental disturbances, thus several existing nonlinear phenomena that must be considered in the control system design are experimentally analyzed. Also, the designed CPI structure is tested and validated through both conducted computer simulations and experimental tests. As demonstrated in the conducted experimental voltage ripple analysis, the voltage ripple amplifications due to the cascaded control structure of CPI impose a limit on the speed control loop bandwidth in CPI, thus its speed control performance can only be improved to a certain degree in practical applications. Therefore, a novel Fuzzy-Linear Quadratic Regulator (F-LQR) is designed in which the error compensation dynamics of the baseline LQR is manipulated through a Fuzzy Logic Controller (FLC) to improve the speed control performance in the transient state while maintaining the optimality in the steady state. Due to its state-feedback type MIMO control structure, improving the speed control performance in the novel F-LQR does not increase the voltage ripples. Also, a geometrical characterization strategy is adopted when shaping the fuzzy mapping of the FLC in which the desired characteristic manipulation in F-LQR is achieved. The designed CPI and F-LQR control systems are comparatively analyzed by conducting experimental tests on the real-world test setup under two compelling scenarios to assess speed control and voltage ripple performances under excessive torque loading and successive reference tracking conditions. The comparative experimental results revealed that the F-LQR outperformed LQR and CPI controllers in terms of both reference tracking and disturbance rejection. Also, a discussion to highlight the pros and cons of these two different control structures is provided from a micromobility traction application point of view. To improve control system robustness and reliability while reducing production costs, two different sensorless speed and position estimation algorithms i.e., Extended State Observer with Phase Locked Loop (ESO-PLL) and back-EMF observer with PLL (EMF-PLL) are developed. A hall-effect sensor-based position estimation strategy is also presented to obtain position information in standstill and critically low-speed operating regions. To achieve reliable and accurate estimations of rotor speed and position under disruptive internal and external disturbances, a proper selection of tuning parameters of the designed observers is also given. The speed and position estimation performances of the designed observers are comparatively analyzed by conducting computer simulations and experimental tests on the real-world test setup. Three different realistic test scenarios are applied in which the abrupt accelerating/ decelerating, successive reference tracking including field weakening region, and excessive torque loading conditions are investigated. The results of the conducted computer simulations and experimental tests revealed the advantages of the EMF-PLL observer over the ESO-PLL and Hall-based methods in all operating regions of the SPMSM.
-
ÖgeHybrid compliance control of collaborative robots(Graduate School, 2022-10-18)Robots are a valuable part of the manufacturing industry for their capability to be precise, consistent, and fast. Therefore, robotics is an essential piece of Industry 4.0. In processes that need high speed/high force robots are preferred, because they are programmed to bring maximum efficiency. Various control schemes are developed for automation of any kind of robot and the control is done via sensors and actuators. The compliant control scheme allows to adjustment of force and torque exerted by the environment onto the robot manipulator during operations. In this thesis, an indirect hybrid position/force controller is developed using a UR10 collaborative robot, for deburring processes. Every deburring process may require specific force control parameters due to variations in workpiece material, form, burr heights, etc. The developed control scheme aims to provide flexibility by making these control parameters adjustable. So that, the same manipulator can be used with different components that require specific deburring parameters. Whereas the built-in force controller in the robot is closed to the end-user. First, the kinematic modeling of general robot structures is explained. UR10 kinematics is shared using Denavit - Hartenberg parameters. Next, dynamic modeling methods are explained which defines the relationship between robot motion, and manipulator forces/torques. Jacobian concept is summarized, because CB3 series Universal Robots do not have a pre-installed F/T transducer at end-effector to measure the forces and torques. The controller is estimating these values on the tool, through Jacobian and motor torque constants at each joint. During testing, noise is observed in the data collected, hence low-pass filters are introduced. It is concluded that the Bessel filter is most suitable due to its characteristics in magnitude and phase responses through experiments. Motion control is explored through various control schemes in free and compliant environments. Compliant schemes are used in applications where force and position are controlled in different directions in a coordinate frame. Hybrid position/force control schemes are explained, and block diagrams are shared. An indirect hybrid compliant control scheme is created to be used with the UR controller. The control scheme includes proportional, integral, and derivative gains that are determined through testing. The control scheme is tested on a flat surface and a curved pipe. Reaction force and displacement results are compared with results from tests with the built-in force controller. In conclusion, comparison results are discussed. Shortcomings of the study and their impact are shared. Future possible improvements are recommended.
-
ÖgePath planning with hybrid use of artificial intelligence algorithms in autonomous mobile vehicles(Graduate School, 2022-06-16)It is aimed to solve the path planning problems of autonomous mobile vehicles with ROS support indoor applications. The genetic algorithm was chosen as a path planning algorithm and applied to the robot. The reason for choosing the genetic algorithm is that there are few studies in the literature evaluating the performance of genetic algorithms in environments with dynamic objects. Genetic algorithms are of stochastic algorithms. Stochastic algorithms have to run a large number of tries to plan an optimal path. A high number of attempts requires good processor performance, otherwise, the planning time of the optimal path may be long in environments with dynamic objects. These trials are selected based on the evaluation criteria. The optimal path in this study is marked on the map with waypoints and the vehicle follows these points. This optimal path, which is revealed when all of the points are passed, is the shortest path between the starting point and the goal point. In order to apply the planner to the autonomous mobile robot, the autonomous system architecture must first be created.