LEE- Kontrol ve Otomasyon Mühendisliği-Yüksek Lisans
Bu koleksiyon için kalıcı URI
Gözat
Son Başvurular
1 - 5 / 50
-
ÖgeFlight control law design for fighter aircraftsexploiting nonlinear control techniques(Graduate School, 2025-07-03)Fighter aircrafts are military aircrafts that are designed to conduct mainly air-to-air missions which require high agility and maneuverability. The high agility and maneuverability are provided by a suitable aerodynamic design and a fly-by-wire flight control system that augments the stability and control characteristics and improves the handling characteristics of the airframe such that the pilot workload for flying the aircraft is minimized. The measure of pilot workload and handling qualities are determined based on quantitative flying and handling quality metrics that are provided in various military standards. These metrics sort of guide a flight control law design to achieve predicted Level 1 handling qualities, therefore flight control law design is based upon these criteria in order to achieve minimal pilot workload. However, satisfying solely handling quality metrics is not sufficient by itself in order to meet airworthiness which requires satisfaction of additional requirements that are related to stability robustness of the flight control loop which are essential for the safety of flight. There are various flight control law approaches utilized in the industry for fighter aircrafts. The first production fly-by-wire fighter aircraft F-16 utilizes classical linear control laws which use pitch rate and angle of attack as feedback variables for the pitch axis. And it uses stability axis roll and yaw rate along with lateral acceleration feedback for the lateral-directional axes. In addition to the linear control law, a nonlinear prefilter called the dual lag filter is used in the roll axis for improved handling qualities. The use of nonlinear components falls back to as early as the first production fly-by-wire fighter aircraft in the 1970's. The Swedish fighter aircraft JAS-39 that conducted its first flight in 1988 also utilizes nonlinear filters in its flight control law to complement its linear control laws. The nonlinear filter that is used in this aircraft is called phase compensating rate limiter and it is used to recover the sudden phase loss that is experienced once the nonlinear control surface rate saturation occurs. The sudden phase loss during control surface rate saturation increases the effective delay of the loop which can cause instability and result in a catastrophic event as experienced with JAS-39. The nonlinear phase compensating rate limiter had solved this issue for this fighter aircraft. EF-2000 developed during the late 1990's and early 2000's has utilized a linear differential PI algorithm along with nonlinear control laws to cope with nonlinear dynamics which cannot be attenuated with linear control laws. The nonlinear control law was composed of a partial dynamic inversion that cancels specific adverse nonlinear dynamics that the linear control law cannot satisfactorily mitigate. Moreover, the state-of-art F-35 utilizes a flight control law that is based on nonlinear dynamic inversion. This solution theoretically allows one to replace the aircraft's dynamics with the desired dynamics however requires highly complex aerodynamic and inertial database of the aircraft in its flight control computer and its performance highly depends on the fidelity of the model implemented in its flight control computer. In this thesis, nonlinear control techniques applied in various industrial flight control law applications are gathered under one application and exploited during the flight control law design of a fighter aircraft for improved flying and handling qualities. A linear flight control law design is conducted based on a multi-stage optimization scheme that is inspired from industrial optimization toolboxes. The dual lag filter of the F-16 and the phase compensating rate limiter used in JAS-39 are re-evaluated and included in the linear design and analysis activities using describing function techniques. The partial dynamic inversion approach method used in EF-2000 are added to the flight control law such that the nonlinear dynamics that cannot be captured during the linearization are compensated using these terms while not altering the linear design. In the first chapter, a brief review of flight control laws used in the industry for several fighter aircrafts are explained. The nonlinear components used in these fighter aircrafts are highlighted and explained for their reasonings. In the second chapter, the derivations of the equations of motion of an aircraft used in the nonlinear six-degree of freedom aircraft simulation model are explained. The common axis types used for the depiction of the aircraft motion are given and dynamic equations for the translational and rotational motion of an aircraft are derived using the transport theorem and Newton's second law. Moreover, Euler's kinematic equation that relate body axis angular rates to Euler angle rates are depicted. The aerodynamic model of the subject fighter aircraft retrieved from NASA Langley wind tunnel tests are provided. The mass and inertial properties of the subject aircraft are given. The environmental model used for the calculation of air density, dynamic pressure and equivalent airspeed is explained. The third order actuator models for each control surface of the subject aircraft with their rate and position saturation values retrieved from the literature along with latencies and noise filters for each control variable are provided. In the third and fourth chapters, trim and linearization of the nonlinear simulation model are explained and stability and control analysis of the bare airframe is conducted using the aerodynamic database and linear aircraft models. Static lateral and longitudinal stability derivatives are inspected with respect to angle of attack for various sideslip angles with leading edge flap and aileron to rudder interconnect gain schedule to determine the maximum safe angle of attack that the aircraft can reach without loss of control or departure. For this, commonly used metrics LCDP, 〖C_n〗_(β,dyn) and deep stall angle of attack are investigated and the smallest angle of attack from these metrics determined the maximum safe angle of attack. After that, rigid body mode natural frequencies and damping ratios of the lateral and longitudinal axes are inspected with respect to equivalent airspeed and altitude to determine linear control law design points. In the fifth chapter, design requirements used in the flight control law design and analysis activities to measure the FHQ level and stability robustness of the closed loop aircraft are explained. The design requirements consist of single loop requirements such as gain and phase margin that are expected to be met according to military standards along with μ analysis of simultaneous broken loop transfer matrices for a more comprehensive stability robustness assessment. The FHQ criteria mostly consist of the military standards, Gibson criteria with an additional criterion to address PIO II issue. In the sixth chapter, the multi-objective optimization scheme used in the design of linear control laws is explained. The multi-objective optimization used in this thesis is mostly inspired from CONDUIT and has multiple stages wherein a certain set of requirements are met. The requirement sets are met in the order of importance regarding the safety of flight. So, in the first stage linear controller parameters are optimized until all of the stability related requirements are met. As soon as stability requirements are satisfied the second stage of the optimization starts wherein FHQ and PIO requirements are met by optimizing the controller parameters. In the following stages, a sum objective is minimized while preserving the feasibility of the earlier results by defining the prior objectives as the constraint function. This way an optimal solution according to the defined objective function can be obtained. A min-max strategy is utilized such that the maximum of the objective functions belonging to the requirement set is used in the optimization process. This way satisfaction of all of the objectives in the requirement set is guaranteed and computational cost is reduced. Each objective function for the requirements is normalized such that an objective score of one corresponds to the Level 1 of its related requirement. This way an objective score of one correspond to the Level 1 value of any requirement normalizing all of the requirements. In the seventh chapter, nonlinear control elements that are composed of nonlinear control laws and nonlinear filters are explained. Inertial coupling phenomenon is experienced under conditions with high angular rates such as loaded roll or high angle of attack maneuvering which can cause loss of control and departure of the aircraft. Linear controller can lack the performance to suppress this and thereby nonlinear control techniques are utilized. The derivation of the nonlinear control law that cancels the inertial coupling terms of the nonlinear aircraft dynamics is conducted performing a partial dynamic inversion. This way only the terms that cause the inertial coupling phenomenon are cancelled leaving the remaining terms to be handled with the linear control laws. In addition, a nonlinear control law that generate the required yaw acceleration to cancel out the gravitational acceleration terms that cause sideslip angle build up is derived. This way additional sideslip angle suppression is attained. Moreover, a nonlinear filter called the dual lag that is used as a prefilter for the roll axis and provides a smooth roll in and abrupt roll out motion for improved roll handling is analyzed using describing function methods. The parameters of this filter are chosen based on an agility metrics study. Also, phase compensating rate limiters for each primary control surfaces are determined and the behavior of a phase compensating rate limiter is explained in the frequency domain using again the describing function method. In the eighth chapter, the linear control law designs for the longitudinal axis are explained. Two distinct control laws are designed with one of them commanding normal acceleration and the other one commanding pitch rate. A stability and control augmentation system (SCAS) that is composed into feedforward and feedback paths is employed. The feedforward path consists of a prefilter that is used to replace or sort of mask the T_(θ_2 ) of the airframe and replace it with a desired T_(θ_(2,des) ) such that the desired FHQ is achieved. The feedforward component places a zero on the forward path such that the integral pole is cancelled and closed loop response resembles that of a classical second order pitch rate response that is important for satisfactory FHQ. Control variable modifications on both normal acceleration and pitch rate are done such that the kinematic coupling between the longitudinal and lateral axes are minimized aiming to achieve a smoother and low order response for each axis that does not alter one another. Objective functions for each optimization stage are tabulated and optimization results are given along with stability and FHQ analysis results. In the ninth chapter, the linear control law design for the lateral axis is explained. A stability and control augmentation system (SCAS) with stability axis roll rate as the command variable is employed as the flight control law. An approximated sideslip angle rate without the use of sideslip angle information is used in the directional axis instead of a stability axis yaw rate for improved sideslip suppression along with Dutch-roll damping augmentation. In addition, a lateral acceleration feedback, that is an alternative to sideslip angle feedback is used in the directional axis for Dutch-roll frequency augmentation and sideslip suppression. An aileron to rudder interconnect is used to minimize sideslip angle build up due to the kinematic relation between angle of attack and sideslip angle. The design of the aileron to rudder interconnect for sideslip minimization and the choice of the location of the sensed lateral acceleration are deeply explained. Again, objective functions for each optimization stage are tabulated and optimization results along with stability and FHQ analysis results are provided. In the tenth chapter, various flight maneuvers are conducted via nonlinear simulations to assess the behavior of the aircraft with both linear and nonlinear control laws. Aileron roll, barrel roll, Herbst J and pull-up push-over maneuvers are conducted to compare the performance of the nonlinear control laws and showcase the improvements with control variable modifications and nonlinear filters. In the last chapter, the study is summarized, analysis results are evaluated and future studies are discussed.
-
ÖgeLSTM-based learning of automata models for production line operations(ITU Graduate School, 2025)Production Lines and Conveying Systems are the staple of modern manufacturing processes. Manufacturing efficiency is directly related to the efficiency of the means of production and conveying. Modelling of manufacturing processes open the door for better structuring and planning the manufacturing processes. A system model allows for the visualisation of the production process and therefore aid in improving production output. Modelling in the industrial context has always been a challenge due to the complexity that comes along with modern manufacturing standards. Conveying systems, representing a major component of production lines, are chosen as the target to model to present an approach applicable in large scale production lines in a simpler format. An automaton is a model representing a system, as would a state-flow diagram, where it can be defined by the inputs, outputs, transitions, states and a mapping function. A turn-key approach that considers the sequential nature of production and conveying processes is considered. Long Short-Term Memory is a pattern recognition Recurrent Neural Network, that is utilised in this study on a simple pneumatic conveying system which transports a wooden block around the system to identift the automata model of its operation. Recurrent Neural Networks (RNNs) capture temporal dependencies through feedback loops, while Long Short-Term Memory (LSTM) networks enhance this capability by using gated mechanisms to effectively learn long-term dependencies. Data from sensor readings are used to train the LSTM in order to output an Automaton that visualises the operation of the pneumatic conveying system, which is controlled by a Programmable Logic Controller. The PLC was utilised to record the sensor readings from the pneumatic conveying system, where each observation was recorded at a predefined sampling frequency. The data used to train the proposed LSTM approach was divided into two groups for training and testing. Additionally, the appropriate cycle length, to narrow down the average cycle length of the pneumatic conveying system, was measured over the course of a large number of trials. The automaton obtained from the proposed LSTM approach is compared with the automaton obtained from OTALA, which is an automaton visualisation algorithm. The outputs of both methodologies resulted in an automaton. The hyperparameters used to obtain a high accuracy automaton, using the proposed LSTM based approach, were obtained through testing and changing each parameter to monitor the effect it had on the accuracy. Furthermore a consistent range of accuracy was maintained throughout the testing therefore it is considered robust. However, the resultant LSTM automaton proves to be a more accurate representation of the conveying system, unlike the one obtained from OTALA.
-
ÖgeExploring player behavior: The role of intrinsic and extrinsic motivations in bartle's taxonomy of player types(ITU Graduate School, 2025)The gaming industry has witnessed great developments over time, with cumulative progress and the leadership of large companies. As players witnessed these developments, their expectations rose and demands arose for features that were very different from a game. With the emergence of these demands and the rise of expectations, game designers faced great challenges in order to meet these demands. Game design is a field that aims to understand player behavior and decision-making mechanisms and to develop mechanics and frameworks accordingly. Moreover, the game design process includes many different parameters and various restrictions, such as technical limitations, gameplay mechanics, strategies created to increase player loyalty as much as possible. Within these different types of limitations and developed techniques, players have been divided into different categories in many different branches, just like games, in order to better meet the expectations of the players. The most accepted of these classifications for a long time has been Richard Bartle's taxonomy of player types (1996). Bartle classified players into 4 different categories: achievers, explorers, killers and socializers. This typology provides an important and deep framework for perceiving the unique expectations and different motivations of the players and making developments accordingly. One of the most important elements of game design is the core gameplay loops that significantly affect the players' experiences. The core gameplay loop consists of challenges, actions and feedback mechanisms that lead the player to certain decisions and try to keep the player connected to the game. Reward is the last link in this loop chain and is also the most important element since it is a response to the effort and labor the player has given. How effective the rewards can vary depending on the type of player playing the game and their individual expectations. Players may have different desires and motivations in the game, so it is one of the important and critical responsibilities of the game developer to respond to this expectation as necessary. Players usually expect a reward in return when they reach a certain point or manage to overcome a challenge. Without a reward, the challenge passed has no meaning. These awards can be given with different methods or frequencies depending on the nature and type of the game. Awards not only affect the game experience but also have a great effect on motivating the player and creating long-term loyalty. The effects of these awards can also vary depending on the player's motivation. Player motivation can be examined under two main headings as intrinsic and extrinsic motivation. While intrinsic motivation can be formed with awards such as witnessing an atmosphere in a game or discovering a new story, extrinsic motivation is formed with more measurable awards such as points and leaderboard rankings. In this case, understanding the connections between Bartle's player typologies and reward systems and creating game experiences that can engage and satisfy the player is of great importance. This study aims to provide a detailed understanding of the game industry and game design strategies, game design processes and their relationship with player behaviors. In addition to theoretical analyses, this study includes a game development study examining players' decision-making mechanisms, different expectations and motivations according to different player types. Players are classified according to Bartle's player typology and players are expected to make decisions when faced with different choices such as increasing their own rank or helping a side character. These decisions are analyzed and the motivations and behaviors of different player types and their approaches to rewards and challenges are examined. The findings of this research are expected to establish a great relationship between player types and reward systems. By analyzing the effects of different reward structures on various player types, the aim is to inform game developers about how to better connect players to the game and enhance their overall experience. This study can help improve reward mechanisms in game design, personalize them, and develop more engaging gaming experiences. Consequently, the research seeks to offer a solid foundation on theoretical game designs and player-centered development.
-
ÖgeEnhancing aircraft avionics safety: Integrating stpa with arp 4761 analysis(ITU Graduate School, 2025)The continuous advancement of engineering technology has significantly influenced society's need for automation in aviation. Modern airplanes have far more software and digital avionics than earlier models, including flight-management systems, digital fly-by-wire actuators, and advanced sensor-fusion networks. This enhancement in capabilities yields operational advantages, improved fuel efficiency, superior economic performance, and increased scheduling flexibility, while concurrently increasing functional complexity. The rapid expansion of codebases into millions of lines, along with the intricate relationships among software, firmware, and hardware, has pushed traditional safety evaluation methodologies to their limits. Authorities like the FAA and EASA do not mandate a singular development process but acknowledge established methods of compliance. SAE ARP 4754 delineates a development assurance framework, while its counterpart SAE ARP 4761 specifies the set of safety assessments that support that framework. While no advisory circular officially references ARP 4761, it is widely embraced across the industry. Classical studies in ARP 4761, Fault Tree Analysis (FTA), Failure Modes and Effects Analysis (FMEA), and associated probabilistic methodologies, were developed for electromechanical systems and are proficient in documenting loss scenarios initiated by component failures. They are not as skilled at identifying dangers arising from software-induced or control-interaction issues that do not originate from a physical defect. Therefore, the thesis argues for the enhancement of the ARP 4761 process through the incorporation of the System-Theoretic Process Analysis (STPA) methodology. Grounded in system theory rather than reliability theory, STPA examines control structures to identify dangerous control actions and causal linkages that conventional methods may neglect. The study hypothesis posits that STPA can reveal supplementary danger situations and formulate preventive requirements, thereby enhancing both vehicle-level and system-level safety artifacts. The study starts with a Functional Hazard Assessment (FHA), which categorizes failure conditions based on their operational impacts and determines the allocation of Development Assurance Levels (DAL). FTA is utilized to allocate quantitative failure-probability objectives; nevertheless, software components are assigned DAL classifications rather than probabilistic budgets, underscoring the discrepancy that STPA addresses. Applying STPA's four-step process subsequent to the FHA produces clear safety limits, hazardous control measures, and loss scenarios. These outputs contribute to Preliminary System Safety Assessment tasks, including FTA refinement and Common Mode Analysis, by offering causal narratives that pure reliability models are unable to deliver. The case study focuses on the Flight Management System (FMS) and its Vertical Navigation (VNAV) function, illustrating the methodology employed. The FMS consolidates route planning, performance calculation, and automated execution, functioning as the operational brain of contemporary planes. VNAV manages climb, cruise, and descent trajectories using sub-functions including profile generation, climb-thrust scheduling, descent management, constraint management, and autopilot/autothrottle integration. By employing both FHA and STPA for these VNAV sub-functions, the study generates a more comprehensive set of safety requirements and demonstrates how STPA reveals interaction-driven dangers that the FHA–FTA cycle would overlook. In conclusion, the integration of STPA with the current ARP 4754/4761 safety workflow provides a more comprehensive assurance strategy for software-intensive aircraft. It maintains the regulatory integrity of current standards while providing engineers with a structured perspective for emerging, non-failure-based risks, aligning with the rapid automation important to contemporary aviation.
-
ÖgeSemi-heuristic optimization techniques performance in robust control systems design(Graduate School, 2025-07-11)Designing a robust controller for systems with parameter uncertainties is a complex and demanding task since they are prone to changes regardless of the external impact forced upon these systems. This change might not only yield poor controller performance, but it can also lead to instability based on the change magnitude. Traditional deterministic control approaches may not provide a convenient solution due to combined computational complexity, and probabilistic approaches may also fall short in providing efficient and satisfactory solutions due to their time-intensive behavior and not guarantee of convergence. To address this challenge, we propose a semi-heuristic approach that works as a mild algorithm, leveraging the advantages of both deterministic and heuristic approaches and overcoming the drawbacks obtained from these two types of control strategies. The proposed semi-heuristic approach exploits the Kharitonov theorem to establish an initial stable controller using system nominal values. Then this controller is used as a starting point for the gradient descent algorithm. The random search technique propagates based on cost function/s implemented for the whole of the uncertainty region. The iterative optimization process by gradient descent incorporates user-defined performance criteria, our approach provides a robust controller with respect to the presence of uncertainties for systems with interval polynomial characteristic equations. We further enhanced the semi-heuristic approach by utilizing a more effective random optimization technique known as the Adaptive Moment Estimation (Adam) optimizer. We applied this proposal to the rotary inverted pendulum system, which is a well-known nonlinear and unstable system in control theory. We managed to demonstrate the efficiency of the semi-heuristic approaches in stabilizing a linearized rotary inverted pendulum model with a parametric approach used for uncertainty representation. Extensive numerical simulations done on the proposed semi-heuristic approaches and the designed model validate the effectiveness of our proposed algorithms for both of the systems considered during this thesis study. Consequently, the semi-heuristic approaches emerge as a moderate solution to our initial problem introduced by the deterministic and stochastic approaches. We further compared the controller results with common control strategies in state-of-the-art and proved their superiority in addressing robust control problems with uncertain parameters.