LEE- Açık Deniz Mühendisliği-Yüksek Lisans
Bu koleksiyon için kalıcı URI
Gözat
Son Başvurular
1 - 2 / 2
-
ÖgeA generalized deep reinforcement learning based controller for heading keeping in waves(Graduate School, 2022-06-21)Reinforcement Learning (RL) is a machine learning method where a learner (the agent) tries to maximize a reward by learning how to act under different environmental circumstances. The agent looks at the state of its environment (through the state vector), takes an action, and then gets a reward and the next state of its environment. The agent improves its action-taking strategy (policy) with every action it experiments with. RL methods have been used for many decision-making problems including control problems with promising results. Unlike many traditional control methods, a model-free RL doesn't need any environment dynamics to operate. This is especially beneficial for problems where the model dynamics are non-linear or not well-known. However, classical controllers are still the most used method of control for maritime applications. Heading-keeping is a maritime control problem where a controller's objective is to keep the heading (yaw) angle of a vehicle constant. Generally speaking, the industry standard is to use traditional feedback controllers such as PID for this problem. This study focuses on designing a generalized RL controller for the heading-keeping problem in waves. The study compares the designed RL controller to a traditional controller in terms of yaw error and rudder usage and observes that the designed RL-based controller performs better than the used traditional controller. The first iterations of the RL agent had many issues. Unlike traditional controllers, the RL agents don't inherently recognize that in an idealized environment they can deal with waves coming from 0 and 180 degrees with almost zero rudder usage. On top of that, the first few developed agents had problems with excessive rudder usage, steady-state error, and overshooting behavior. All of these problems have been solved in the final iteration of the RL agent. Instead of just explaining the final agent, the thesis starts off with a weak RL agent and explains how it can be improved iteratively. This way the thesis explains how one might approach the problem of developing an RL-based controller. The first section focuses on giving a rough summary of RL and the problem case, explains the purpose of the thesis, then talks about previous work over marine movement control in literature. Some detailed information about the used tools and simulation environment is also given here. The second section introduces LQR controllers and designs an LQR controller for the heading keeping problem. The third section explains RL in-depth to lay the foundation for the upcoming sections. The fourth section starts with a naively designed simple RL agent and iteratively improves it. In each iteration of development, the agent is compared to the designed LQR controller, its weaknesses are analyzed, and the improvements for the next iteration are determined. The fifth section summarizes the previous sections, explains the contributions of the thesis, and discusses possible future work.
-
ÖgeNumerical modelling of waves and current acting on piles(Graduate School, 2023-06-21)Calculating the applied forces of pile structures exposed to waves and currents is an important issue in the offshore industry. Because these constructs are costly, they must be carefully analyzed before evaluating all results. For designers, physical models can be cost and time constraining while helping them optimize their designs. Therefore, the use of numerical models provides a significant advantage. While numerical models offer flexibility in the design of complex structures, their accuracy must be compatible with physical experiments. In this study, the effects of a pile on a flat ground subjected to waves were numerically modeled. These numerical models were solved using the open-source computational fluid dynamics code called REEF3D. In addition, a mesh-convergence study was performed to determine the optimal mesh size. Numerical models of wave and current study samples were constructed using the optimal mesh size and the results were compared with experimental data and analytical results. It was observed that the numerical model was compatible with the experimental results in pile structures affected by waves and currents. First, two test scenarios were created for validation. In the first of these scenarios, only current effects were observed. These current effects were compared with the results of previous physical experiments. Some observation points were used in this comparison. These observation points were placed at certain distances from the single pile perpendicular to the base. The velocity information obtained from the observation points was compared with the numerical model results modeled with REEF3D and velocity profiles were obtained. These velocity profiles were found to be compatible with the results of the physical experiment and the numerical model. In the second verification scenario, only wave effects were observed. Wave forces acting on a fixed diameter cylinder were investigated. Numerical models were established and the results were obtained. These results were compared with the Morison equation results and the results were found to be consistent. After both verification scenarios were made, it was revealed that both wave and current forces can be modeled correctly with the REEF3D program. After this step, the forces acting on four different pile types in eight different scenarios were examined. To understand the diffraction effects, half of the eight scenarios were constructed with parameters where diffraction effects were considered important. At the beginning of these parameters is the change of the pile diameter. In the first of four different piles in eight scenarios, a pile diameter with a constant diameter and such that diffraction effects can be ignored was chosen. A new pile of this diameter was then created to create a cone of appropriate proportions. Wave and current forces acting on this new cone-shaped pile were calculated. The forces acting on both the fixed diameter pile and the cone shaped pile were compared with the results of the Morison and Beji equations and it was seen that reasonable results were obtained. On the other hand, pile scenarios with relatively larger diameters were created and fixed diameter and cone-shaped piles were remodeled in these pile scenarios. Evaluation of all results was re-evaluated and reasonable results were found. A separate mesh-convergence study was performed for all cases. For the mesh convergence study, a coarse mesh size was chosen and gradually reduced. It was desired to see whether it converged to such a value or not. The mesh created by the meshconvergence study were optimized and it was predicted that it could yield reasonable results. At the end of the studies, the compared results were presented both in tabular form and graphically. These results, obtained thanks to open-source tools, showed that we can produce solutions to engineering problems with REEF3D. In the final case, the forces generated by different waves and currents acting on the pile placed on a flat base were analyzed. Thanks to this study, wave or current forces acting on the piles were calculated, and further studies revealed that both wave and current forces could be calculated by computational fluid dynamics method. The use of numerical models has been a turning point in the analysis of pile structures exposed to waves and currents. Due to the limitations and costs of traditional physical experiments, the flexibility and cost-effectiveness of numerical models provide a great advantage. Open-source computational fluid dynamics codes such as REEF3D allow engineers to further examine the behavior of pile structures. This study demonstrates the potential of REEF3D for the analysis of pile structures exposed to waves and currents. It has been shown that the forces acting on the pile structures can be accurately calculated using numerical models. This helps designers make better decisions about the safety and durability of piles. In addition, the validation scenarios and analyzes performed in the study showed that the numerical models are compatible with the experimental data and analytical results. This supports the reliability and accuracy of the numerical models. The results of the study emphasize that numerical models are an important tool in the design of pile structures exposed to waves and currents. Using these models, designers can better understand, optimize, and safely design the behavior of piles. In the future, more advanced numerical models are expected to be developed and used for the analysis of more complex pile structures and different environmental conditions. Moreover, it will be possible to obtain more comprehensive and accurate results by integrating different open-source tools and computational methods.