LEE- Kontrol ve Otomasyon Mühendisliği-Yüksek Lisans
Bu koleksiyon için kalıcı URI
Gözat
Son Başvurular
1 - 5 / 43
-
ÖgeA new anti-windup strategy for fractional order PI controllers(ITU Graduate School, 2025)Fractional calculus originated in 1695 and was later expanded by various mathematicians. Among the most commonly used fractional-order definitions are the operators proposed by Riemann-Liouville, Grünwald-Letnikov, and Caputo. These formulations extend classical differentiation and integration to non-integer orders, allowing for more flexible and detailed mathematical modeling. In control engineering, fractional-order controllers offer enhanced design flexibility, enabling the development of controllers with improved performance and robustness characteristics. Implementing fractional-order controllers requires certain approximations. One of the most widely adopted methods is proposed by Oustaloup. This aims to construct an integer-order transfer function that closely emulates the frequency response of a fractional-order operator. In control system implementations, various physical constraints must be considered, one of the most significant being actuator limits. These limitations can lead to a phenomenon known as integral windup, which occurs when the control signal produced by the controller exceeds the actuator's physical capabilities. The integrator continues accumulating error, resulting in an excessively large control signal. This mismatch degrades the system response by increasing overshoot, prolonging settling time, and potentially leading to instability. To address this issue, numerous anti-windup techniques have been proposed in the literature. This study focuses on a comparison of four leading and widely used methods: back calculation, automatic reset configuration (ARC), ARC with fractional filter and fractional integral order reset. While most of these techniques are formulated based on integer-order PID structures, some have been adapted or developed for fractional-order controllers. However, these often fall short of leveraging the flexibility offered by fractional calculus. This thesis, firstly, focused on key factors contributing to integral windup. Among these, saturation limits define the physical constraints of the system, while controller parameters, including the proportional gain, integral gain, and integral order, also play a critical role. Each factor influencing the growth of the control signal can worsen windup; however, the effect of the integrator, particularly the order of integrator, is found to be especially significant and is thoroughly investigated in this thesis. To evaluate anti-windup performance, existing methods are tested on first-order plus dead time (FOPDT) systems, which are widely used in control applications due to their ability to capture the dominant behavior of various dynamic systems, facilitate system identification, and approximate higher-order models. The proposed stategy is designed and generalized specifically for such systems, using their characteristic parameters. Two key performance criteria are defined for windup mitigation: the tracking error between the reference and system output, and the magnitude of saturation error. Based on these, a method is proposed in which the fractional integral order is adapted online. By dynamically updating the integral order, the proposed structure improves system performance while suppressing integral windup. The update rule for the integral order relies on two criteria: the tracking error and the saturation error. The variation of these metrics is analyzed in relation to the FOPDT system parameters: the process gain, time constant, and time delay. These relationships are then incorporated into the update rule, allowing the proposed strategy to be expressed as a generalized formulation for FOPDT systems. In conclusion, the proposed strategy has been tested alongside existing approaches on various FOPDT systems. The comparison is based on four performance metrics: overshoot, rise time, settling time, and peak time. The results demonstrate that the proposed approach provides a notable advantage in overshoot suppression and generally yields a more stable system response compared to the other methods.
-
ÖgeSavaş uçakları için dört boyutlu bir uçuş yönetim sisteminin tasarımı(Lisansüstü Eğitim Enstitüsü, 2025-06-25)Havacılık ve uzay teknolojisindeki ilerlemeler hava muharebelerinin dinamiklerini sürekli olarak değiştirmektedir. Özellikle 5. nesil olarak adlandırılan günümüzün gelişmiş savaş uçakları, hava-hava ve hava-yer görevlerinin tek platformda yapılmasını sağlayan gelişmiş araçlardır. Bu araçlar, çeşitli sensör üniteleri, görev üniteleri ve farklı özelliklerde radarlardan oluşan gelişmiş aviyonik sistemlerle donatılmıştır. İnsansız hava araçlarıyla tümleşik görev kabiliyeti, akıllı mühimmatların kokpitten yönetimine imkan tanıyan sistemlerle birlikte birer uçan bilgisayar haline gelen bu çok rollü hava araçları, oldukça kabiliyetli güç çarpanlarıdır. Ancak bu gelişmiş, kompleks sistemler pilota oldukça yüksek miktarda veri beslemektedir. Savaş uçağı pilotları hava aracını sürerken bir yandan da görev ile ilgili birçok sistemi de yönetmek zorundadır. Bu kompleks muharebe ortamı pilotun iş yükünü ciddi miktarda artırmakta, pilotun farkındalığını düşürmekte ve en önemlisi de görev başarımını zorlaştırmaktadır. Bu tez çalışması, modern savaş uçakları için pilotun iş yükünü azaltacak ve pilotun farkındalığını artıracak 4 boyutlu bir uçuş yönetim sisteminin tasarımını ve simülasyon sonuçlarını göstermeyi hedeflemektedir. Tez çalışmasının birinci bölümünde yapılan çalışmanın amacı, literatürdeki 4 boyutlu uçuş yönetim sistemlerinin ve savaş uçaklarında kullanılan uçuş yönetim sistemlerinin kabiliyetlerinin incelenmesi ile başlamaktadır. Literatürdeki eksikliğin tanımlanması ve bu çalışmanın hedefi olan hipotez açıklanmaktadır. Tez çalışmasının ikinci bölümünde açık kaynaklı parametreler ile oluşturulmuş F-16 savaş uçağı benzetim modeli anlatılmaktadır. Bu model, çalışma kapsamında tasarımı hedeflenen uçuş yönetim algoritmasının savaş uçağı dinamiklerine sahip bir sistem ile gösteriminin yapılması amacıyla kullanılmıştır. Tez çalışmasının üçüncü bölümünde hava aracı benzetim modelinin otonom navigasyon yapabilmesi için tasarlanan navigasyon, otopilot ve kararlılık artırma sistemlerinin tasarım mimarisi anlatılmaktadır. Kontrolcü tasarımında klasik PID kontrolcüler kullanılmıştır. Tez çalışmasının dördüncü bölümünde tezin ana konusu olan 4 boyutlu uçuş yönetim sisteminin rota yönetim algoritmasının kabiliyetleri, planlama algoritması kapsamında yapılan dikey ve yatay planlama hesaplamaları, zaman ve yakıt planlamaları, yürütme algoritması kapsamında yapılan navigasyon hesaplamaları anlatılmıştır. Tez çalışmasının beşinci bölümünde F-16 savaş uçağı benzetim modelinin örnek bir uçuş rotası kullanılarak, 4 boyutlu uçuş kontrol sistemiyle otonom navigasyon sonuçları gösterilmiştir. Örnek rota uçuş rotası için planlama ve yürütme algoritmalarının çıktıları vurgulanmıştır. Sonuç olarak bu tez, 4 boyutlu bir navigasyon sisteminin savaş uçaklarına nasıl uyarlanacağını göstermiştir. Savaş uçakları için 4 boyutlu bir uçuş yönetim algoritması konsepti tasarlanmış, UYS simülasyon modeli geliştirilmiş ve farklı durumlar için simülasyon sonuçları verilmiştir. Önerilen sistem, bir uçuş rotasının 4 boyutlu olarak planlanmasını ve otonom olarak icrasını sağlamaktadır. Sonuçlar incelendiğinde, bu algoritmanın savaş uçaklarında pilotun iş yükünü azaltacağı, pilotun farkındalığını artıracağı ve böylece görev başarımının artacağı sonucuna varılmaktadır.
-
ÖgeImmersion and invariance disturbance observer based discrete time vector control of BLDC machines(Graduate School, 2025-06-26)Nowadays, with the use of electric motors in every field of daily life, the amount of electricity consumed by these electric motors has increased and today's studies are focused on making both the control and driver designs of these electric motors more efficient and able to operate at higher power levels. Today, electric motors are not only limited to industry or production processes but also started to be used with the main traction drive motors in electric vehicles, and the efficiency of these motors has a significant effect on the total range of the vehicles, and more efficient motors have also started to be preferred in these vehicles. The relationship between efficiency and price is not a design criterion that can be ignored. Along with the increasing demand for efficiency in electric vehicles, the types of motors used also vary due to price reduction policies. Synchronous motors are mostly preferred in electric vehicles. Synchronous motors are motor types in which the magnetic field produced by the stator and the rotor rotate at the same speed. However, this is not the case in asynchronous motors, and the rotor speed follows behind the magnetic field produced by the stator. When these motors are compared in terms of efficiency and price, the following arguments emerge. Among asynchronous motors, the induction motor has become one of the first preferred motors in the electric vehicle sector. This type of motor has short-circuited conductive plates on the rotor. The working principle of this motor is that the magnetic field produced by the stator induces a current on these conductive plates and this induced current ensures that the rotor rotates. Induction motors provide a more affordable price advantage than synchronous motors due to their structure, only having windings on the stator and the simple structure on the rotor, but since there is no structure to produce a magnetic field on the rotor, they have lower power density and therefore less efficiency than synchronous motors. When focused on the PMSM or BLDC motors, which is the most used in the sector among synchronous motors, there are permanent magnets on the rotor parts of these motors. These magnets constantly produce a fixed magnetic field. Since the magnetic field is produced by the rotor, energy does not need to be spent to re-generate the entire magnetic field by the stator, so these machines are more efficient than induction machines. Thanks to the magnets in the rotor structures, these motors also have higher power density. PMSM and BLDC motors are similar to each other in their basic structures. Both motors have stator windings and permanent magnets on their rotors, and these magnets produce the necessary magnetic field. The difference between these two motors is the difference in the waveforms of their back EMFs. While the back EMF in PMSM machines is sinusoidal, this back EMF is square wave in BLDC motors. This thesis study focuses on developing a discrete-time vector control method based on immersion and invariance disturbance observer for BLDC motors. In this context, firstly the structures of BLDC motors are considered and electrical and mechanical equations are derived from these structures. Apart from electrical and mechanical equations, the electromechanical torque equation that acts as a bridge between these two structures and connects the two systems is derived. Using these equations, a state space equation is derived for the BLDC motor and then an immersion and invariance based disturbance or in other words, a disturbance torque observer is developed based on this derived state space equation. The main purpose of the immersion and invariance based disturbance observer is to observe and estimate the disturbance torque effects experienced by the system on the shaft of the BLDC machine. These disturbance torque factors are inherently dependent on variables that cannot be measured or are very difficult to measure. Examples of the parameters that produce this immeasurable distorting torque include friction in the motor's balls, unequal conditions experienced by the motor during the production phase, the ideal unequal phase resistances and inductances of the motor, and the unequal heating of the machine due to an unequal ventilation system. Later, the vector control structure, which is the most preferred method for three-phase synchronous motors, was studied and this control method was tested in a simulation environment. The main idea in the vector control method is to be able to control the magnetic field production and torque production in brushed DC machines separately. While brushed DC machines achieve this by having separate windings in their stators and rotors, this study cannot be achieved in BLDC machines because there are permanent magnets in the rotor section. At this point, the vector control method converts the three-phase rotating currents into two constant currents at ninety degrees to each other using the transformations called Clarke/Park. These currents are called d-q axis currents, and while the d current controls the magnetic field produced, the q current controls the torque produced by the machine. In this way, the components that produce the torque and magnetic field are separated from each other as in brushed DC machines. The most preferred external control loops in the vector control method are known as torque control and speed control. These controls actually send control signals to the d and a current controllers as reference values as muscle layers with the speed control being the outermost. In this study, the speed control structure was preferred instead of torque control. The disturbing torque estimated by the immersion and invariance-based observer was then connected to the vector control structure as a pre-fed term, thus eliminating the effects of the disturbing torque on the system and achieving a more robust control. Here, the estimated torque was added to the reference torque, which is the output of the speed controller, and fed to the system. The speed, d and q axis current controllers, an inverter model, a BLDC machine model and a load model required for vector control were modeled in the Simulink environment. The reason for including the inverter model in this modeling is that the system has an inverter in real-life operation. Therefore, it was aimed at creating a modeling environment that exhibits more realistic dynamics. All components modeled in this modeling environment were modeled modularly. In other words, each of these models was developed as a separate function and the components such as motor resistance, inductance, and number of poles they needed to operate were provided parametrically from outside. Thanks to this parametric structure, BLDC motors with different structures and parameters can be changed quickly without making major changes to the system, which increased the mentioned parametric structure of the system. The developed models were first tested one by one in the Simulink environment and their operation was verified, then all models were connected to each other and the entire system was first operated in open loop and then in closed loop. After verifying that the proposed method works, a test environment containing an isolation transformer, DC input capacitor, inverter, BLDC motor, controller, induction motor and load was designed so that this method could be tested in real life. Attention was paid to keeping the cost of this test environment low. The hardware and software design of the inverter used in this test environment was made within the scope of this thesis. At the same time, the system design for the test environment and the implementation of this test environment were provided. After this design was realized, the six-step control method, which is widely preferred in BLDC motors due to its simplicity, was first implemented and the commissioning and verification studies of the designed test environment were carried out. After the commissioning studies of the test environment, the proposed method was tested on this test environment. The tests carried out in the simulation environment were compared with the tests carried out on a real engine in the test environment and the results were presented with the study of the proposed method.
-
ÖgeComparison of quaternion-based orientation estimation methods using 9-dof marg sensors(Graduate School, 2025-06-12)Orientation estimation using inertial and magnetic sensors has become a fundamental capability across robotics, autonomous vehicles, and wearable systems. Accurate and low-latency orientation tracking enables balance control, motion planning, and sensor fusion in environments where external references like GNSS or visual markers may be unreliable or unavailable. Among various representation methods, quaternions are widely adopted for their singularity-free, computationally efficient handling of 3D rotations. This thesis presents a comprehensive evaluation of quaternion-based orientation estimation algorithms using a nine-degrees-of-freedom (9-DOF) inertial and magnetic sensor also known as a magnetic, angular rate, and gravity (MARG) sensor, under various motion scenarios. These filters ranging from simple complementary filters to more advanced solutions like Madgwick, ESKF (6-DOF and 9-DOF), XKF3i, and the modular Versatile Quaternion-Based Filter (VQF), were tested for their ability to fuse gyroscope, accelerometer, and magnetometer data into consistent orientation estimates. Quaternions were used to avoid singularities and provide stable performance in dynamic multi-axis movements. Experimental validation was conducted using both free-hand and robot-guided motion. In hand trials, the IMU was manually moved through varied rotational patterns. For robot trials, a UR3 robotic arm followed pre-programmed trajectories, with the IMU rigidly mounted on its end-effector. Reference orientation was derived from an OptiTrack motion capture system and, in robot trials, forward kinematics. Two data acquisition pipelines were used: a manual logging approach for early trials and a ROS-based synchronized recording system for later robot trials, capturing IMU, motion capture, and joint states. Filter performance was assessed by comparing estimated orientations against reference data, using both quaternion-based angular errors and Euler angle deviations. Quantitative evaluation relied on Root Mean Square Error (RMSE) and normalized Integrated Time-weighted Squared Error (ITSE) metrics, which capture both instantaneous accuracy and cumulative drift over time. Results varied across motion types: filters performed similarly in slow or repetitive motion but diverged under fast or complex dynamics. Magnetometer integration generally improved yaw accuracy but introduced sensitivity to environmental disturbances. Simpler filters like Madgwick and the complementary filter offered fast computation and easy deployment, while Kalman-based methods required careful tuning but were more robust. XKF3i provided strong results but lacked transparency. The VQF filter, on the other hand, delivered the most consistent and stable performance on average - though not in all scenarios-, benefiting from its modular correction of tilt, heading, and bias. The findings highlight that no single filter excels universally; instead, trade-offs depend on application needs such as accuracy, computational load, and sensitivity to noise. Future work could include tests in magnetically disturbed environments, improved calibration and tuning automation, and extensions toward translational motion estimation or multi-IMU fusion on articulated platforms. Overall, this thesis offers a practical framework for filter selection and tuning, supporting the development of resilient orientation tracking providing a structured experimental basis for selecting and tuning orientation filters in robotics, wearable sensing, and human–machine systems.
-
ÖgePerformance comparison of guidance algorithmsfor aerial vehicles: a software-in-the-loop based approach(Graduate School, 2025-06-26)This study aims to comparatively examine the guidance algorithms used for mini cruise missiles, based on the increasing importance of cost-effective and precision guidance weapon systems in today's battlefields. Mini cruise missiles have a wide range of use due to their low cost, rapid deployment capabilities and precision engagement capacities, exemplified by platforms such as Baykar Kemankeş-1 and ASUW Spear3. While such missiles are particularly effective in missions requiring high precision against point targets, they are also of critical importance in the modern battlefield with their ability to engage moving targets. Cruise missiles generally stand out with their ability to stay in the air for long periods, to conduct reconnaissance and surveillance, and to attack targets with high precision. However, in cases where targets are not fixed and move at variable speeds and directions, realistically evaluating the performance of these vehicles requires complex, costly and risky testing processes. Therefore, it is of great importance to create realistic simulation environments and perform performance analysis in these environments. Within the scope of this thesis, a realistic Software-In-The-Loop (SITL) environment was created by integrating different software tools and projects. SITL is a method that allows the developed software to be tested in a simulation environment like real hardware before using real hardware, thus reducing development costs and accelerating the design process. Missile Datcom software was preferred for fast and effective determination of aerodynamic parameters; this software plays a critical role in the design phase by providing missile aerodynamic coefficients quickly and accurately. JSBSim flight dynamics modeling software was used for realistic modeling of missile dynamics. JSBSim enables the creation of realistic flight scenarios with its ability to model flight dynamics in a detailed and flexible manner. The open source Ardupilot platform, which provides SITL support, was used as the autopilot system. Ardupilot enables fast and reliable integration of different guidance algorithms thanks to its flexible architecture. For the visualization of the simulation, Unreal Engine-based Antoinette Project was used, allowing users to visually follow the test scenarios in real time and in detail. In the study, target vehicles were tested on different motion dynamics and scenarios. In this context, constant speed and constant acceleration motion scenarios of the targets were created, and the direction of motion was determined on the earth's axis. The basic algorithm used, Proportional Navigation Guidance (PNG), is a classic and reliable method that produces the necessary maneuver commands by taking into account the change in the approach angle of the missile to the target. The Augmented PNG (APNG) algorithm, which was developed to increase the effectiveness of this algorithm, provides more precise tracking by taking into account the acceleration of the target and sudden changes in its motion. The Modified PNG (MPNG) algorithm, by adding the lateral deviation term to the guidance algorithm, allows the missile to initially turn slightly to the side of the target rather than in its direction, allowing it to maintain its precision strike capability in the terminal phase. True PNG (TPNG), on the other hand, provides higher accuracy guidance capability by using the target's real motion direction and speed information more effectively. Through the scenarios created, each algorithm's performance criteria such as sensitivity, target acquisition time, energy efficiency and adaptation capabilities to different target movement conditions were analyzed and compared in detail. Thanks to this study, valuable data was obtained on the effectiveness of the guidance algorithms of mini cruise missiles under different operational conditions, providing an important infrastructure for systems to be developed in the future.