研究方向

This study employed a dynamic mesh model to comprehensively examine the performance of the E1619 propeller considering the influence of the free surface effect. The turbulence flows around the propeller surface were captured using a large eddy simulation (LES) model, while the water surface was generated by utilizing the volume of fluid (VOF) method. The forward motion of the propeller was generated by implementing a dynamic mesh model and the velocity motion was defined using an user-defined function (UDF). The simulation results were validated through a series of open water tests conducted in the NCKU towing tank, achieving a high success rate in attaining total uncertainties of approximately 6% or less. The discretization analysis was employed via a simultaneous grid and time refinement approach. The simulation results in the case of a single propeller and propeller with strut under a multi-phase condition were used to analyze the hydrodynamic performance as well as the flow structure around the propeller, leading to influential parameters based on the achieved level of accuracy. It was demonstrated that the case of a propeller with a strut demonstrated superior predictions compared to single propellers, with thrust coefficient and propeller efficiency errors below 6% for single phase and 5% for multi-phase. Conversely, the case of a single propeller in a single phase exhibited more accurate torque coefficient estimations with errors below 2%.

The hydrodynamic coefficient is an important indicator for evaluating the fluid performance of underwater vehicles, and the Planar Motion Mechanism (PMM) test is one of the main testing methods. This study aims to simulate the Sting-Supported Planar Motion Mechanism (SPMM) test and investigate the effects of different motions, including oblique tests, pure surge, pure swaying, pure yawing, pure heaving, pure pitching, and pivot rolling, on the hydrodynamic coefficients of the full-scale DARPA SUBOFF model. By adjusting the period, amplitude, and different yawing velocities, the study examines the influence on the hydrodynamic coefficients. Additionally, this study compares the hydrodynamic coefficients obtained by the David Taylor Research Center (DTRC) for the DARPA SUBOFF model to determine the numerical settings that best match real experiments. In this study, the full appendages DARPA SUBOFF model is utilized, and the Computational Fluid Dynamics (CFD) simulations are conducted using a combination of the LES model and Dynamic Mesh Method for the PMM test. To simulate the real flow field conditions, the simulations are performed using a captive model test setup, representing the model's towing in the flow field, and establish the computational domain based on the actual dimensions of the towing tank at National Cheng Kung University. In terms of the grid aspect, the study employs a discretization error analysis to ensure the convergence of grid size and time step. Subsequently, the numerical simulations are performed to obtain the corresponding hydrodynamic coefficients.

流體動力係數（Hydrodynamic Coefficient）是評估水下載具流體性能的重要指標，而平面運動機構（Planar Motion Mechanism, PMM）試驗是其中主要的測試方法之一。本研究旨在模擬以後插式平面運動機構（Sting-Supported Planar Motion Mechanism, SPMM）試驗，探討全尺度的DARPA SUBOFF模型在斜航試驗（Oblique Test）、純縱移（Pure Surge）、純橫移（Pure Sway）、純橫擺（Pure Yaw）、純起伏（Pure Heave）、純縱搖（Pure Pitch）和橫搖（Pivot Roll）的姿態下，通過調整週期和振幅以及不同平擺速度對流體動力係數的影響。同時，本研究將比對大衛泰勒實驗室（David Taylor Research Center, DTRC）對DARPA SUBOFF模型所測得的流體動力係數，以找出最貼近真實實驗的數值設定。本研究使用DARPA SUBOFF全附屬物模型，結合LES模式和動態網格技術，進行斜航試驗和平面運動機構試驗的計算流體力學（Computational Fluid Dynamics, CFD）模擬。為了模擬真實的流場情況，本研究以拘束模型試驗（Captive Model Test）的形式進行模擬，呈現模型在流場中的拖航姿態。以國立成功大學的拖航水槽之實際尺寸為參考，建立了本研究的計算流域。在網格方面，本研究使用離散化誤差（Discretization Error）分析確定網格尺寸和時間步長的收斂性，而後經過數值運算，以獲得相應的流體動力係數。

In this study, a computational fluid dynamics (CFD) model was developed to simulate the aerodynamic performance of the National Renewable Energy Laboratory (NREL) offshore 5-MW baseline wind turbine with single rotor and full wind turbine .Using statistical methods, the relation between pitch angle and performance when the speed is higher than the rated wind speed was analyzed; furthermore, other published data were compiled to establish the functional equations of power, thrust with various inflow wind speeds, and pitch angles. Then, the transient mode is used to simulate the instantaneous interaction of the blade when passing through the tower, and finally the atmospheric boundary layer is imported to discuss the aerodynamic performance of the wind turbine in the actual wind field under the northeast and southwest monsoon conditions, respectively. In addition, according to shape optimization based on parametric modeling, the two -dimensional cross -section of the wind turbine blade can be defined through a metasurface approach, and the three -dimensional surface modeling of the wind turbine blade, nacelle , and tower is completed using the nonuniform rational B -splines (NURBS) interpolator. In terms of aerodynamic simulation, the unstructured polygon mesh was used herein to discretize the space and simulate the highly curved and twisted surfaces of the blade. In terms of aerodynamic simulation, the unstructured polygon mesh was used herein to discretize the space and simulate the highly curved and twisted surfaces of the blade. In this study, the compact and accurate mesh form obtained through a grid independence test will be used to analyze the distribution of the pressure coefficient, shear stress coefficient, and limited streamline on the blade surface at various inflow wind speeds and pitch angles to understand the differences between different turbulence models and the causes of power and thrust attenuation at high inflow wind speeds. In addition, the transient mode additionally analyzes the time series of instantaneous air flow, torque, thrust, and tower drag as the blade passes through the tower; energy amplitude spectrum, and wind profile of the flow field to obtain tower effects and atmospheric boundary layers effects on wind turbine aerodynamic performance.

本研究建立一套計算流體力學 (Computational Fluid Dynamics, CFD)模式以模擬NREL 5-MW baseline turbine在單轉子(Single Rotor)及全風機(Full Wind Turbine)情況下空氣動力表現(Aerodynamic Performance)，並利用統計方法分析於穩態模式(Steady State)下，不同額定風速(Rated Wind Speed)下之槳距角 (Pitch Angle)與有效功率及推力比間的關係，並歸納其他公開發表之數據以分別建立功率(Power)、推力(Thrust)與額定風速及槳距角之函數方程式(Functional Equation)。接著，利用暫態模式(Transient State)模擬轉子於通過塔架時的瞬時交互作用，最終導入大氣邊界層(Atmospheric Boundary Layer, ABL) 探討風機分別在東北、西南季風條件下，實際風場中的氣動力學表現。另外，根據參數化建模(Parametric Modeling)進行形狀優化的方式，風機葉片可經由超表面方法(Metasurface Approach)定義二維截面，並由非均勻有理B雲規曲線插值器(Non-Uniform Rational B-Splines (NURBS) Interpolator)完成風機葉片三維表面建模。在空氣動力模擬部分，本研究利用非結構化多面體網格(Polygon Mesh)離散化空間，藉此模擬高度彎曲和葉片的扭曲表面。經由網格獨立性測試(Grid Independence Test)所獲得之兼具經濟性及準確性的網格形式，後續將用以分析在不同風速及槳距角的風機尾流(Wake Flow)分佈、葉片表面的壓力係數(Pressure Coefficient)、剪應力係數(Shear Stress Coefficient)及極限流線(Limited Streamline)之分佈情形，以了解不同紊流模式(Turbulence Model)間的差異，及其造成在高額定風速下，有效功率及推力比降低的原因。此外，暫態模式還另外分析葉片通過塔架之間的瞬時空氣流動及扭矩、推力以及塔架阻力之時序列及能量振幅譜，還有流場風速剖面，以此獲得塔架效應(Tower Shadow Effect)以及大氣邊界層對風機空氣動力學性能的影響。

In this study, the aero-hydro-elastic-mooring coupled dynamics of the NREL offshore 5-MW baseline wind turbine supported by the Offshore Code Comparison Collaboration Continuation (OC4) DeepCwind semisubmersible floating platform were investigated by considering the effects of suddenly broken mooring behaviors. According to the Norwegian Petroleum Directorate wind spectrum and the Joint North Sea Wave Project spectrum, the wind and wave loads are selected as inputs of the Cummins time-domain equation. To acquire hydrodynamic coefficients, the three-dimensional panel method was adopted to solve potential-flow diffraction and radiation problems of hydrodynamic forces in the frequency domain. The nonlinear viscous drag was estimated using the quadratic damping matrix instead of Morison’s element. In addition, the quadratic transfer function matrix was used to calculate the slow-drift force. Compared to the multi-segmented quasi-static mooring model, the lump-mass method is thought of as a dynamic mooring model and can solve real-time motion on the platform motion. Once a single or a pair of mooring lines are intentionally disconnected from the floating platform at a certain time, the transient responses of mooring line tensions and turbine performance will be evaluated. Eventually, the drift trajectories and power performances of OC4 DeepCwind semisubmersible floating offshore wind turbine under different conditions of broken mooring lines can be approximately estimated in the wind farm. 

本研究分析OC4 DeepCwind半潛式浮式風機在不規則風浪條件下的動態響應，環境條件基於NPD風譜與JONSWAP波譜，考慮二階波浪激盪力並作用於繫泊模型。本研究使用三維小板法求解線性化勢流繞射與輻射問題，計算出流體動力係數。為了引入非線性黏性阻力的影響，本研究採用二次阻尼矩陣取代莫里森公式。本研究使用二次轉換函數(QTF) ，藉由堆積質量法計算慢漂移，並藉由研究纜繩模型對運動響應的影響。在動態模型中採用堆積質量法，而準靜態模型則採用優化過的多段準靜態(MSQS)模型。最後，本研究採用風-波浪-繫泊的模擬，對OC4 DeepCwind半潛式浮式風機繫纜發生單條與雙條斷裂時的性能進行分析。本研究在不同斷纜情況下，分別考慮風機運轉中與關閉的狀況，針對風機漂移的軌跡進行計算，分析結果可作為未來浮式風場在初步設計階段下的參考依據。

This study aims to design and implement an "underwater image monitoring system" for autonomous underwater vehicle(AUV). The proposed AUV navigates with a stereo vision imaging system composed of dual-lens cameras, also uses Faster R-CNN object detection of underwater targets, and the bounding box of object detection as the control of the rudder plate angle and propeller speed of the AUV. The depth map is predicted by a semi-global block matching (SGBM) based on stereo matching algorithm using images from the navigation records, combined with deep Q-network (DQN) which is deep reinforcement learning method to focus the agent's attention on the target region in the disparity map.The agent decides the SGBM cost parameter, optimizes the disparity map by calculating the reward function, and finally uses the disparity map for depth calculation to obtain the 3D point cloud of the underwater target.

The combination of DQN stereo matching solves the problem that SGBM could only use the try-and-error method to adjust the parameters to get a better disparity map. Finally, this study investigates the stability of AUV in dynamic tracking experiments at different wave heights and the effect of using image information for control, conducted data analysis and 3D reconstruction of images. It was found that the attitude change of AUV under the wave force would affect the position of the target object in the images, resulting in a large error in the size of the object in the stereo reconstruction.

本研究旨於設計與實現「水下影像監測系統」於自主式水下無人載具(Autonomous Underwater Vehicle, AUV)。提出AUV航行中使用雙鏡頭相機組成的立體視覺影像系統，並用Faster R-CNN檢測水下目標物，以物件檢測之邊界框作為AUV控制航向舵板角度與螺槳轉速之控制。使用航行紀錄之影像以半全域區塊匹配(Semi-Global Block Matching, SGBM)為主的立體匹配演算法來預測深度圖，並結合Deep Q-Network(DQN)深度強化學習方法使代理人將注意力集中在視差圖中目標物區域上決策SGBM代價參數，經獎勵函數的計算對視差圖進行優化，最後利用該視差圖進行深度計算，獲得水下目標物之立體點雲。結合DQN之立體匹配解決SGBM以往只能使用試誤法調整參數得到較理想的視差圖的問題。最後，本研究探討AUV於不同波高動態追蹤實驗之穩定性與利用影像資訊進行控制之效果，並進行數據分析與影像立體重建，發現在波浪外力作用下AUV之姿態變化將影響目標物於影像中位置，導致立體重建之物體尺寸誤差較大。

This paper develops an Adaptive Neuro-Fuzzy Inference Systems (ANFIS) combined with Reinforcement Learning (RL) for Autonomous Underwater Vehicle (AUV) path following control simulation and experiment. RL is the process of learning to perform a mission correctly by interacting with a dynamic environment to obtain reward values. Therefore, this paper develops an AUV maneuvering simulation system as an environment for RL to train AUV to achieve path following. The maneuvering simulation system is based on six-degree-of-freedom motion maneuvering equations to simulate the maneuvering performance of AUV during task execution. After the Q-learning training in RL, the Q-table of the optimal control policy (horizontal and vertical rudder) for path following in different environments is established.

The environment input variables and the corresponding output states of the Q-table are trained by ANFIS. ANFIS is a combination of Neural Network  principles and fuzzy inference, which can infer fuzzy membership function and fuzzy rules by itself. The horizontal and vertical rudder with different states of Q-table are synthesized as the controllers of vertical and horizontal planes in this paper. In addition, three different Deep Reinforcement Learning (DRL) vertical plane controllers combining Q-learning algorithm and ANFIS are designed according to three different reward functions. Finally, through AUV path following experiments to discuss the effect of the designed controllers under different wave conditions.

本論文以自適應類神經模糊推論系統(Adaptive Neuro-Fuzzy Inference Systems，簡稱ANFIS)結合強化學習(Reinforcement Learning, RL)應用於自主式水下無人載具(Autonomous Underwater Vehicle, AUV)的路徑追蹤控制模擬與實驗。強化學習(RL)是透過與動態環境互動獲取對應獎勵值來學習正確地執行一項任務，因此本文開發AUV操縱模擬系統做為強化學習(RL)之環境，訓練AUV達成路徑追蹤，操縱模擬系統以潛體六自由度運動操縱方程式為基礎，模擬AUV在執行任務時的操縱性能表現。透過強化學習(RL)中Q-learning訓練後，建立路徑追蹤在不同環境下所對應的最佳控制策略(水平舵板與垂直舵板)之Q-table。將Q-table建立之環境輸入變數與對應之輸出狀態透過ANFIS進行訓練，其中ANFIS是結合類神經網絡(Neural Network，NN)原理及模糊推論，自行推論模糊歸屬函數(Membership Function, MF)與模糊規則，最後以擬合出根據Q-table下所有不同狀態之水平舵板與垂直舵板，作為本文垂直平面與水平平面之控制器。另外，根據獎勵函數的變化設計三種不同結合Q-learning演算法與ANFIS之深度強化學習(Deep Reinforcement Learning, DRL)垂直平面控制器，並進行AUV下潛路徑追蹤實驗，並探討不同波浪條件下，所設計之控制器的效果。

The purpose of this study is to establish a two-way fully-coupled numerical model that can be used as an aero-hydro-elastic simulation of offshore wind turbines for analyzing their natural frequencies excited by environmental forces such as wind and waves, as well as the influence of the mode shape on the dynamic response. According to the parametric modeling for shape optimization of the blades, two-dimensional cross sections can be defined by the metasurface approach and the three-dimensional surface modeling can be completed by using non-uniform rational B-splines (NURBS) interpolator. At the beginning, both the aerodynamic and hydrodynamic simulations will be conducted based on the open source computational fluid dynamics (CFD) software OpenFOAM. In addition, the structural dynamics will be simulated based on the open source multi-body dynamics software MBDyn. Subsequently, the user-defined dynamic motion library in OpenFOAM is utilized to treat the complex CFD mesh motion in an aero-hydro-elastic simulation. Initially, the proposed numerical model will be applied to explore the hydroelastic effect of slamming wave loads on two different types of offshore wind turbine foundations (i.e. monopile and jacket types) after including the towers. Meanwhile, the modal analysis is used to investigate the dynamic amplification of shear stress and overturning moment corresponding to each mode. In addition, the aeroelasticity of the rotor under wind shear effect and the fluid-structure interaction between the aerodynamics and the tower will evaluate the influence on the structure deformation through modal analysis. Eventually, this study will conduct a fully coupled analysis of the aero-hydro-elasticity for the offshore wind turbine affected by winds and waves simultaneously, compare the differences caused by decoupling analysis of winds and waves separately, and then comprehend the influence of wind turbine design on the dynamic response.

本研究目的在於建立一套可作為離岸風機之氣動水彈性模擬的雙向全耦合數值模式，以分析離岸風機受到風和波浪等環境外力激發後的自然頻率，及其相關的模態對動態響應的影響。根據參數化建模進行形狀優化的方式，風機葉片可經由超表面方法定義二維截面，並由非均勻有理B雲規曲線插值器完成風機葉片三維表面建模。首先，本研究將以開源計算流體力學軟體OpenFOAM為基礎進行空氣及水動力的模擬，並以開源多體動力學軟體MBDyn為基礎進行結構動力的模擬，再利用使用者自定義的網格運動函式庫來處理氣動水彈性模擬中複雜的 CFD 網格運動。本研究初步將開發的數值模式應用於探討砰擊波荷載對於兩種不同型式的離岸風機基樁(單樁式及套管式)在搭配塔柱後所造成的水彈性效應，並進行模態分析以研究各模態對剪應力及傾覆力矩所造成的動力放大。另外，風機轉子在風切變下的氣動彈性，及其空氣動力在轉動過程中與塔柱的流固耦合作用，將可透過模態分析進行評估對結構變形的影響。最後，本研究將在風、波浪同時作用下對離岸風機之氣動水彈性進行全耦合分析，並比較單獨對風和波浪進行解耦分析時所造成的差異，進而了解在風機設計改變下對動態響應的影響。 

In this study, the CFD method is used to carry out numerical simulation of the interaction between the offshore wind turbine foundation and a series of regular waves. The Reynolds Average Navier-Stokes (RANS) equation is combined with the RNG k-ε turbulence closure to explore the transient phenomena of run-ups around the monopile foundation. The finite volume law is a numerical simulation for discrete solutions. The finite volume method can be used to solve the two-phase flow of air and water on the free surface in the numerical wave tank. This study investigates the evolution of regular waves propagating over the slope and its influence on the run-up heights of offshore wind turbine foundation by using the piston-type wave generator. In order to demonstrate the accuracy of wave generation, a series of regular wave cases conducted in the super wave flume at Tainan Hydraulic Laboratory of the National Cheng Kung University were considered to perform numerical simulation. In addition, the wave-maker theory was used to simulate the real function of the piston wave-maker. The numerical wave-maker is mainly based on the dynamic layering method for simulating the actual interaction between the wave-maker and the fluids. Finally, in addition to using the run-up equation to estimate the vertical water elevation caused by the interaction of the wave and the structure, this study also uses the fast Fourier transform, Hilbert-Huang transform and empirical mode decomposition method to analyze the evolution of waves over the sloping bed.

In this study, hydrodynamic derivatives of a full-scale Defense Advanced Research Projects Agency (DARPA) SUBOFF model were calculated by computational fluid dynamics (CFD) modeling of horizontal planar motion mechanism (HPMM) tests. The unsteady volume of fluid (VOF)-based Reynolds Averaged Navier-Stokes (RANS) solver was performed on oscillation motions of the model by adding a comprehensive dynamic mesh strategy. In order to simulate motions of pure surging, pure swaying, pure yawing, yaw motion with a constant drift angle, pivot rolling motion, as well as the straight-line motion with drift angles in the HPMM tests, the dynamic mesh strategy consisting of the smoothing method, the dynamic layering method and the local remeshing method has been well established. Consequently, the hydrodynamic derivatives analyzed in the yaw mode, the pitch mode and the roll mode for both the bare hull (configuration 3) and the bare hull with sail and rudders (configuration 8) would be compared to the experimental data of configuration 3 model. The simulation results of configuration 3 are basically in good agreement with the experimental data; the ones of configuration 8 provide for the complete database of linear and non-linear hydrodynamic coefficients.

本研究旨於設計與實現「潛航控管系統」(Diving Control System, DCS)於自主式水下載具(Autonomous Underwater Vehicle, AUV)，基於載具深度計設計、強健型模糊控制器決策優化設計與即時載具航行姿態閉迴路控制，完成動態測試並進行實驗數據分析。深度計設計使用絕對壓力型式液壓計做為感測元件，採用進行壓力深度換算，構成本研究使用於AUV之深度感測輸入裝置。潛航控管系統控制載具之平擺(Yaw)與俯仰(Pitch)角度，由載具配置的姿態航向參考系統(Attitude and Heading Reference System, AHRS)及深度計即時獲取載具當前之運動姿態向量與入水深度值，執行載具潛航控管演算程序獲得AUV運動控制之垂直舵與水平舵角度值。本研究之潛航控管系統採用強健型模糊控制器進行模糊決策優化設計，並且在實際試驗中與一般未經優化之模糊控制器控制表現進行比較。最後，本研究完成AUV潛航控管系統整合動態測試並進行實驗數據分析發現，經採用決策優化之強健型模糊控制器在精準深度控制及穩定姿態皆優於一般未經優化之模糊控制器。

關鍵字：自主式水下載具(AUV)；模糊控制(Fuzzy Control)；深度控制；航行姿態控制

In order to simulate horizontal planar motion mechanism (HPMM) tests of an autonomous underwater vehicle (AUV) in the towing tank, the research presented here attempts to develop a dynamic mesh strategy based on the Reynolds Average Navier-Stokes (RANS) solver to reproduce the various motions of the HPMM. For each mode of reproduction, the installation mode is divided into yaw mode, pitch mode, and roll mode. Regarding each installation mode, there are driven straight-line motion with constant drift angles, pure surging motion, pure yawing motion, and pure yawing motion with a constant drift angle. The simulation results were verified by conducting a series of HPMM tests in the towing tank of National Cheng Kung University. The dynamic mesh strategy, which consists of the boundary layer smoothing method, the dynamic layering method and the size function-based local remeshing method, was employed to process the mesh update as a result of the AUV motion in the numerical simulation. Since the numerical results are quite consistent with the experimental data, it can be proved that the proposed dynamic mesh strategy is suitable for the simulation of the HPMM test in the towing tank.

This study used computational fluid dynamics (CFD) to simulate breaking waves induced by sloping beds and to analyzes the impact of these waves on an offshore wind turbine (OWT) foundation. A Reynolds-Averaged Navier-Stokes (RANS) equation was established to numerical-ly analyze the water surface of two-phase fluids by using a volume of fluid method. The Boussinesq approach was used to determine the Reynolds stresses for developing a Re-Normalisation Group (RNG) 𝑘 − 𝜀 turbulence model. The cut-cell method was used to generate grids for the monopile geometry and the computational domain. The results of the numerical model for the wave elevation, wave load, and pressure on the foundation structure were validated by comparing them with the experimental results obtained by Irschik (2002) at the Large Wave Channel, Hannover, Germany. Next, the wave elevation and wave load subject to the surf similarity parameter 𝜉0 on the OWT foundation were examined. By using empirical mode decomposition (EMD) and the Hilbert–Huang transform (HHT), the total wave load forces were decomposed into the quasi-static force and slamming force. Furthermore, the wave load analysis would be summarized by considering the slamming force coefficient, rising time, duration time, and impulse. The normalized run-up height of the wave around the OWT foundation was also examined .

Keywords: Breaking wave, Wave load, Computational fluid dynamics (CFD), Offshore wind turbine foundation. 

本研究利用計算流體力學(Computational Fluid Dynamics, CFD)的數值方法對流體力學的控制方程式進行求解與流場模擬，針對離岸風機單樁式基樁與三頻波之間的非線性波群(Non-linear Wave Group)交互作用影響來探討。利用有限體積法(Finite Volume Method, FVM)做為離散求解的數值模擬，數值波浪水槽(Numerical Wave Tank, NWT)中自由液面部分採用體積分率法(Volume of Fraction, VOF)解決空氣與水之兩相流問題；流體與基樁壁面之黏性流問題則採用雷諾平均納維-斯托克斯方程式(RANS)搭配RNG k-ε紊流模型(RNG k-ε Turbulent Model)求解基樁受水動力作用時的暫態情形。在數值水槽模型的網格處理上，對時間步長與網格大小相對關係的庫朗數(Courant Number)進行侷限以增進模擬計算時的穩定與收斂性，因此針對自由液面交界範圍採用體積加密的方式固定其大小，同時也將無因次壁面函數(Non-Dimensional Wall Function) y^+考慮在內，對基樁壁面邊界層處增加使用邊界層網格設定(Inflation)生成六面體網格；並拘束其第一層厚度以得到更為準確的結果，藉此能夠計算出紊流邊界層內的黏性流域對基樁之影響；最後也在水槽的後方採用數值消波(Numerical Beach)的設定，讓波浪在通過基樁一段距離後進行消散以防止回波對結果造成影響。此外在波浪條件的部分，為了了解實際海況的非線性波群作用與波浪調變(Wave Modulation)所造成的影響，本文針對單樁式基樁在三種典型非線性波列之演變進行觀察探討，分別是單一頻率規則波、兩相等振幅之雙頻波、一主頻兩副頻之三頻波的不同條件下之溯升高度位置情形，並導入無因次化調變因子(δ ̂)分析其與基樁周圍之溯升情形。在受力探討的部分，本研究採用除了以Morison的半經驗公式推估流體作用的水平力外，快速傅立葉轉換(Fast Fourier Transform, FFT)搭配Hilbert-Huang Transform和經驗模態分解法(Empirical Mode Decomposition, EMD)來進行能譜時域分析將受力得以分解出準靜態力、動態力與砰擊力(Slamming Force)，使得在波浪荷載的探討與上有較為精確的結果，也有助於日後的離岸風電海洋工程的開發風險影響評估。 

本研究的目的為探討本實驗室自行開發的自主式無人水下載具(Autonomous Underwater Vehicles, AUV)，經由計算流體力學(Computer Fluid Dynamics, CFD)的方法來探討AUV在全附屬物(Bare Hull with Full Appendages)下流場中的情況。數值模擬的紊流模型採用SST k-ω 模型(Shear Stress Transport K-ω model)作為本次的紊流模型，並採用SIMPLEC (Semi-Implicit Method for Pressure-Linked Equations Consistent) 為求解器之設置條件，並參考DAPA SUBOFF的實驗規劃進行斜航試驗(Oblique Towing Test)。分別進行多種入流角(β)的模擬，分析AUV所承受的受力、力矩變化，同時觀察AUV在不同入流角下的流場差異性、壓力分佈、流場變化、壓力係數和摩擦係數分佈。 

本研究目的為開發一具可同時提供人為操作與自主航行的潛艦自航模型 (Submarine free-running model)，並導入全功能操縱及壓載艙系統調整裝置。以Symmechanics 模式將艙體及舵板、帆罩之幾何特性匯入 MATLAB 來建立控制系統。本自航模型的艉舵控制系統在設計上為通過考慮 X 型舵的特性，使用 PID 控制器設計了一個 X 型舵的控制模組來執行舵板控制與航向校正，並透過舵板之流體動力系數來分析艉部的X型舵之間控制面的舵效應，並以計算出時域下的六度自由運動轉換為 X 型舵之舵角，最後再與目前較為常見之十字舵比較其效能之優劣。 

本研究的主要目的著重在視覺辨識及立體影像測深技術之開發，並應用在自主式水下無人載具影像導航系統之建構。在視覺辨識技術上結合霍夫轉換法與光流演算法作為偵測動態影像之直線特徵與移動速度；在立體影像測深技術則是利用Harris角點檢測法推估目標物之距離。經由搭載於艏艙的廣角鏡頭與雙眼視覺裝置所截取的影像作為輸入源，再透過Simscape Multibody所建立的模擬環境，可控制艉艙的後驅動系統(包含水平、垂直舵板及推進器)。在待測目標物的靜態測試上中比較不同亮度、濁度下的分析結果。最後，待測目標物的動態測試結果經結合模糊控制器，也可用以輸出目標物在不同移動速度、距離下，舵板的轉動角度、角速度與推進器的即時反應。

本研究擬以MATLAB/SIMULINK為基礎，建構一套用以推估浮動式風機平台動態反應的模組化模擬系統。本系統主要包含以線性勢能流論為基礎的邊界元素法求解器，及以康明斯時域方程式為基礎的流體動力模擬器，並搭配可視化軟體ParaView進行波浪與風機浮台之交互作用。在本論文中將比較幾種典型的浮式風機平台在波浪作用下的動態反應，及其受準靜態錨鍊系統的影響程度。根據不同剛度矩陣的設定，可用來探討在規則或不規則波海況下三種不同型式錨鍊系統的動態特性。經由反應振幅因子的分析可得知各型式風機浮台所對應的自然頻率，及確保所設計風機的安全性與穩定性。

In order to enhance the efficiency and safety of outgoing ships, this study pro-poses a ship weather routing system that combines the Three-Dimensional Modified Isochrone Method (3DMI) with the Particle Swarm Optimizing (PSO) method. The system is essentially constructed with four modules, including a ship-motion module, an ocean-environmental module, a navigation module, and a routing-optimization module. By means of data obtained from the weather forecasting database, the 6-DOF motion responses of a ship as well as the efficiency performance can be simulated dynamically. All the candidate routes for different purposes are determined ac-cording to two reference sailings, i.e. the Great Circle (GC) sailing and the Ocean-Current (OC) sailing, respectively. Through analyzing the voluntary and involuntary speed loss of voyages, different ship trajectories as well as performances are presented in our system. Eventually, the costs related to sailing distances, passage time, and fuel consumption are discussed in this study. Keywords: Weather Routing System, PSO, Ocean-Current Routes 

本研究以計算流體力學 (Computational Fluid Dynamics, CFD) 方法進行模擬，著重於離岸風機基樁和雙頻波交互作用之水動力探討。數值模擬方面，將以有限體積法 (Finite Volume Method, FVM) 做為離散求解工具；波浪之自由液面處理方面，將以體積分率 (Volume of Fraction, VOF) 方法處理水與空氣之兩相流問題；風機基樁受波浪作用下之黏性流問題，將以RANS方程式搭配RNG k-ε紊流模型 (RNG k-ε turbulent model) 求解基樁受水動力作用時的情形，並使用邊界層網格設定 (inflation)，生成細長六面體網格於基樁周圍之邊界層進行網格細化處理，藉以計算邊界層內之黏性流域對基樁之影響；在消波方面，採用數值消波設定 (numerical beach)，使波浪於通過基樁後進行消散。此外，本文將探討不同型式基樁在不同頻率差情況下之最大與最小溯升高度位置，將以無因次化調變頻率分析基樁周圍之溯升情形，最後探討雙頻波所造成之受力影響。 由模擬結果得知，單樁式、重力式、三角管式和套管式基樁相較之下，週期差對單樁式、重力式、三角管式基樁之衝擊係數影響較大，反觀套管式基樁，週期差之影響則是較小。 關鍵字:非線性波、溯升高度、波力、計算流體力學、固定式離岸風機基樁

The purpose of the present study was to discuss the wave run-up heights and wave impact forces induced by nonlinear wave groups. The Finite Volume Method (FVM) is a discretization technique for solving partial differential equations implemented in Computational Fluid Dynamics (CFD). The Volume of Fraction (VOF) method is suggested as a free surface modeling technique to track and locate free surfaces. The hydrodynamic simulation of the wave-structure interaction was conducted using a Reynolds-averaged Navier-Stokes (RANS) solver for incompressible fluid flows and employing an Re-Normalization Group (RNG) k-ε turbulent closure. In order to solve the viscous flow around the foundation regions, an inflation scheme was used to generate the smaller hexahedral meshes and refine them. To avoid wave reflection, numerical beach treatment was employed. Furthermore, the maximum and minimum wave run-up heights around the foundations for different wave steepness are analyzed by the normalized parameter of frequency modulation . Finally, the wave impact force is also discussed and summarized qualitatively and quantitatively. By means of our CFD model, it’s obviously found that for the jacket foundation period difference did not have a great influence on the value of slamming coefficient in comparison with the monopile, gravity-based and tripod foundations. Keywords: Nonlinear Wave, Wave Run-up, Wave-load, CFD, Offshore Wind Turbine Foundations

本研究主要著重在水下載具加上不同附屬物之流體動力實驗與分析，實驗方面使用平面運動機構(Planar Motion Mechanism, PMM)，數據分析則是使用帶通濾波(Band Pass)將雜訊去除。模型以DARPA SUBOFF為基礎，為了搭配成功大學系統系地下實驗水槽進行實驗改為1/2縮尺之模型，附屬物的部分包含帆罩(Sail)與艉舵(Rudders)，實驗模型配置為裸船(Bare Hull)、裸船加上單獨帆罩(Bare Hull With Sail)、裸船加上艉舵(Bare Hull With Rudders)與全附屬物(Bare Hull With Full Appendages)。PMM機構實驗包含靜態實驗與動態實驗兩種，靜態實驗為斜航試驗(Drift Angle Test)，藉由改變船體角度進而得到不同漂流角度下的縱向受力、側向受力與平擺扭矩之變化情形；動態實驗主要是藉由PMM機構上橫移馬達與平擺馬達來達到縱移實驗(Surging Test)、橫移實驗(Swaying Test)、平擺試驗(Yawing Test)與平擺時驗加上漂流角(Yawing Test With Drift Angle)。縱移實驗與橫移實驗目的為量測船體之縱向與橫向附加質量；平擺時驗與平擺時驗加上漂流角則是探討角速度與角加速度對船體受力與平擺扭矩間的關係。此外，也使用帶通濾波來濾除實驗中台車震動與馬達震動等雜訊，使實驗數據最佳化，並且與DTRC實驗結果做比較。 關鍵詞：平面運動機構(PMM)試驗、操縱性、水下載具、六度運動、帶通濾波

The main purpose of this study is to obtain the hydrodynamic coefficient of four types of 1/2 scale DARPA SUBOFF models, including bare hull, bare hull with sail, bare hull with rudders, and bare hull with all appendages, by using PMM experiments. The PMM experiments include both a static test and dynamic tests, i.e. a sway test, a surge test, a yaw test, and a yaw test with constant drift angle. After obtaining the raw data, a band pass is used to filter high frequency vibrations. Finally, a Fourier transform is used to obtain the hydrodynamic coefficients. The results indicate that the drift angle test hydrodynamic coefficients are similar to those from the DTRC (the David Taylor Research Center), and the dynamic tests are obviously effected by the vibration of the towing carriage and motor. In contrast, the static test has good accuracy, and it is found that the dynamic test data must be analyzed with a band pass, or the vibration of the towing carriage or motor will affect the results. Key words: PMM; submerged vehicle; maneuvering; band-pass filter

本研究以計算流體力學 (Computational Fluid Dynamics, CFD) 方法進行模擬，探討離岸風機基樁和非線性波浪交互作用下之水動力機制。在數值方面，將以有限體積法 (Finite Volume Method, FVM) 做為離散求解工具；而在波浪之自由液面處理方面，則以體積分率 (Volume of Fraction, VOF) 方法處理水與空氣之兩相流問題；風機基樁受波浪作用下之黏性流問題，可用RANS方程式搭配RNG k-ε紊流模型 (RNG k-ε turbulent model) 來求解紊流之影響，並使用inflation設定，生成六面體網格於基樁周圍之邊界層以進行網格細化處理，藉以計算邊界層內之黏性流域對基樁之影響。此外，本文將探討三角管式基樁周圍在不同波浪尖銳度情況下之最大與最小溯升高度位置，將以速度停滯頭理論 (Velocity Stagnation Head Theory) 分析三角管式基樁周圍之溯升情形，最後探討波浪荷載變化情形。

This research focuses on the hydrodynamic interaction between an offshore wind turbine foundation and nonlinear waves, using the CFD model to simulate wave loads and wave run-ups. The FVM is a discretization method for numerical simulation. The VOF is a free surface modeling technique to track and locate the free surface. The hydrodynamic simulation was conducted by using a RANS solver for incompressible fluid flows and employing RNG k-ε turbulent closure. Then, the employment of inflation scheme would generate the hexahedral meshes and refine the meshes of boundary layer of the tripod. The effects of the viscous flow region in the boundary layer around the tripod would be calculated. Furthermore, the maximum and minimum wave run-up heights around the tripod in different wave steepness would be analyzed by applying velocity stagnation head theory. Finally, the dynamics of wave loads would be discussed

由於大型離岸風力發電場之開發已是當今世界各國發展海洋再生能源的主要途徑，有鑒於此，本研究將以計算流體力學 (Computational Fluid Dynamics, CFD) 軟體，發展一套波浪條件模擬方法，並以RANS方程式搭配k-ε紊流模型作為求解器，模擬三種不同固定式離岸風力發電機基樁，受到不同波浪條件影響下，單一風機基樁受外力作用的影響，並探討基樁受波力作用時所產生的溯升高度。 本研究主要著重在水動力的探討，因此在數值方法上，將以有限體積法 (Finite Volume Method) 做為離散求解工具；在波浪條件的自由液面設定上，則是使用體積分率 (Volume of Fraction, VOF) 方法處理水與空氣之兩相流問題；在探討風機基樁受波浪作用下之黏性流問題上，則使用非穩態的標準K-ε紊流模型求解基樁受水動力作用時的情形。此外，也探討風機基樁周圍最大與最小溯升高度位置，以及分析單樁式、三腳管式以及重力式三種不同基樁型式，受不同波浪條件影響下的波浪荷載情形與溯升高度變化。 經模擬分析後發現，在波浪尖銳度較高、週期較短的波浪條件下，三種不同種類的基樁皆有較高的溯升高度，但基樁所受到的波力卻較小；反之，波浪尖銳度較小、週期較長的波浪條件下，雖然溯升高度較小 ，但波力卻會對基樁造成較大的波浪荷載。 本文首先探討計算流體力學的理論基礎、模擬環境的建構與準確度驗證後，再詳細分析不同波浪條件對三種風機基樁的受力、溯升等交互關係。 關鍵詞： 計算流體力學、離岸風力發電機基樁、溯升、波浪荷載、體積分率法。

A hydrodynamic simulation of wave run-up heights and wave loads on three types of wind turbine foundations, i.e. monopile, gravity-based and tripod support structures, was conducted using a RANS solver for incompressible fluid flows, employing k-ε turbulent closure. The present CFD model based on the commercial software, FLUENT, is provided to calculate the maximum wave heights and wave loads that the foundations may experience during the wave propagation in Stokes second order wave theory. The wave run-up on different designs of wind turbine foundations is discussed by the breakwater effect, which would be the dominant factor deciding the minimum air-gap requirement of working platform. Thus, a series of numerical experiments with different cases of wave steepness ka as well as scattering parameter kA were conducted by comparing the effects of wave characteristics on these support structures. The results indicate that the difference among the maximum normalized run-up heights of these support structures is smaller for lower wave steepness than those for high wave steepness. In contrast, it is shown that the difference among the wave loads of these foundations is larger for lower wave steepness than those for higher wave steepness. Key words: CFD, Wave Run-Up, Offshore Wind Turbine, Foundation, RANS.