http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/163 Fri, 21 Aug 2026 09:51:39 GMT 2026-08-21T09:51:39Z http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/16699 標題: ＮＲ車用網路的安全群播; Reliable Groupcast For NR V2X 作者: Yan-Teng Kuo; 郭彥滕 摘要: 隨著自動駕駛汽車的發展，高可靠性和低延遲的車輛通訊技術變得越來越重要。第三代合作夥伴計劃（3GPP）特別討論新無線（NR）下的邊緣連結（SL）傳輸，目的是滿足低延遲，高可靠性和高流通量的車聯網（V2X）要求。與僅支持廣播方案的長期演進技術（LTE）V2X相比，NR V2X還支持兩種方案：單播和群播。群播方案允許發射機一次與同一群中的多個接收機進行通訊。在本論文中，我們呈現了NR V2X模式2的盲重傳。為了保持傳輸的可靠性，車輛群的大小需要被限制，以避免由於非直視性（NLOS）傳輸而引起的錯誤。並且，為了提高傳輸的可靠性，本文提出了一種優化群接受率的重傳方法，可以使發射機在同一群中分配一個重傳裝置進行重傳。我們設計了一個重傳裝置選擇問題，旨在最大程度地提高NR V2X模式2群播的群接收率（GPRR）。模擬的結果顯示，所提出的重傳方法從GPRR、分組間接收時間（PIR）和可靠性限制下的車輛群大小方面改善了車聯網群播。; With the development of autonomous vehicles, the high reliability and low latency vehicular communication technologies has become more important. The Third Generation Partnership Project (3GPP) specifies the sidelink (SL) transmission based on New Radio (NR) structure, which aims to meet the the low latency, high reliability and high throughout vehicle-to-everything (V2X) requirements. Compared with Long Term Evolution (LTE) V2X which only supports broadcast use cases, NR V2X supports two more scenarios: unicast and groupcast. The groupcast operation allows the transmitter to communicate with multiple receivers in same group at once. In this paper, we present the blind retranmission operation for NR V2X mode 2. To maintain the transmission reliability, the group size for vehicles is restricted to avoid errors due to Non-line-of sight (NLOS) transmission. In order to improve the transmission reliability, this paper proposes a optimized group packet reception ratio retransmission, which allows transmitter to assign a retransmission UE in same group to perform the retransmission. We formulate a retransmission UE selection problem, which aims to maximize the group packet reception ratio (GPRR) for NR V2X mode 2 groupcast. The simulation results show that the proposed retransmission method improves the V2X groupcast in terms of GPRR, packet inter-reception time (PIR) and the available group size. Wed, 01 Jan 2020 00:00:00 GMT http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/16699 2020-01-01T00:00:00Z http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/47422 標題: 鼠籠式風力發電機串聯動態電壓調整器之設計; Design of a Dynamic Voltage Regulator for a Squirrel-Cage Induction Generator 作者: Shiang-Cong Chen; 陳翔琮 摘要: 由於節能減碳的需求，全球正快速發展與綠色能源相關之技術，「風」為一天然、無污染之資源，風力發電的技術與理論便成為綠色能源發展中最重要的方向與目標之一。 風力發電所使用的電機種類型式眾多，其中又以鼠籠式感應機價格較為低廉，可大量量產並安裝在風力資源豐富之地區。但由於鼠籠式感應機缺乏對無效電力控制的能力，當系統發生負載加載、切載或故障時，無法負擔支撐起系統的責任，風力不穩定時，感應機便成為整個系統中電力潮流的汙染源，影響電力品質。 為了改善上述缺點，本論文在鼠籠式感應機上串聯動態電壓調整器，控制發電機流入系統匯流排之電力潮流，改善電力品質。並在系統發生加載、切載或故障時，進行適當的補償，穩定整個系統電壓。 本論文利用MATLAB○R/Simulink/SimPowerSystems軟體模擬電力系統負載變動以及風速變動時，發電機及負載端電壓。由模擬結果發現動態電壓調整器能有效穩定發電機及負載電壓。諧波失真也都能符合台電諧波管制標準。; To Reduce emission, much effort has been devoted to the development of green energy techniques. Among the various types of green energy generations, wind energy conversion is studied in thesis. Squirrel-cage induction generators have been widely employed for wind energy conversion due to their low costs and high reliability. However, the squirrel-cage induction generators are not capable of providing voltage regulation due to the lack of rotor excitation. To improve the reactive power support and voltage regulation of a squirrel-cage induction generator, a dynamic voltage regulator（DVR） is designed in this thesis. Through the generation or absorption of reactive power, the DVR is capable of stabilizing the generator terminal voltage and the load voltage when the system is subject to load changes or wind speed variations. The effectiveness of the proposed DVR is demonstrated by computer simulations using MATLAB○R/Simulink/SimPowerSystems. Fri, 01 Jan 2010 00:00:00 GMT http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/47422 2010-01-01T00:00:00Z http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/74913 標題: 點對特徵物體姿態估測結合多視角投票方案用於工業機器人隨機取物; Pose Estimation Using Point Pair Features and Combining Multi-View Voting Scheme for Industrial Robotic Random Bin-Picking Applications 作者: Yao-Jia Kuo; 郭曜嘉 摘要: 本篇論文提出了一種基於視覺的機器人隨機取物，該系統可以對物體進行三維姿態估測。提出的辨識演算法可以整合多視角的資訊進而對估測的姿態進行修正，也提出了一個最佳視角估測的策略進而去減少感測或視角所造成的不確定性。 本篇論文提出的辨識系統是基於點對特徵的方法，此方法使用的哈希的技巧和一種有效率投票的機制達成快速的姿態估測。除此之外，此系統使用了一個驗證函數產生一個分數去檢測估測的姿態。如果估測的姿態是錯誤的，本篇論文提出了一個多視角的方案去修復錯誤的估測。這個方案利用了在點對特徵方法上和賀夫相似的的投票機制，從多視角的資訊將會被整合進而去增加系統的穩定性。 再者，為了減少感測的不確定性和環境的不確定性，本篇論文也提出了一個sensor planning的策略去決定下一個最好的相機姿態去觀測環境。這個決策定義了一個成本函數考慮了下個視角所能得到的視野，同時考量了移動的效率和機器人的工作空間，而且這個決策也有逃離區域資訊的能力。 實驗結果首先展示了當多視角的資訊被整合時，提出的多視角的方案可以得到收斂的結果，而且也可以增進驗證函數產生出的分數。再者，實驗也展示了最佳視角估測可以被用來預測下一個含資訊量最多的視角，同時考量了效率和安全性。最後，辨識系統將會被使用在機器人隨機取物的應用上，藉由一個普遍的六軸機械手臂和配備一個相機。; This thesis presents a practical vision-based robotic bin-picking system that performs three-dimensional pose estimation of the object. The proposed recognition algorithm can integrate the multi-view information to refine the estimated pose, and a sensor planning strategy is also proposed to predict the next best view to reduce the uncertain caused by sensors or viewpoints. The proposed recognition system is based on point pair features (PPF) using a hashing technique and an efficient voting scheme to achieve fast pose estimation. In addition, the system uses a verification function to generate a score to check the estimated pose. If the generated hypothesis is not correct, a multi-view scheme is proposed to fix the wrong estimation. The scheme utilizes the Hough-like voting property in PPF, and the multi-view information can be integrated to increase the robustness of the system. Moreover, a sensor planning strategy is also proposed to decide the next best pose of vision sensors to observe the environment for reducing sensing uncertainty or environmental uncertainty. A cost function is defined for the strategy which considers the expected visibility-gain in the next view in the meanwhile ensuring the efficiency in movement, and the workspace constraint of the robot. The strategy also has the ability to escape from the local information. The experimental results first demonstrate the multi-view scheme can get a convergent result to increase robustness when the information from multi-view is integrated. The scheme can also improve the score generated by the verification function. Moreover, the experiment also shows that the next best view strategy can be used to predict the next best viewpoint which contains the most information and considering the efficiency and safety simultaneously. In last, the recognition system will be used for the robotic bin-picking system by a commonly industrial six-degree-of-freedom mounted on an RGB-D camera. Tue, 01 Jan 2019 00:00:00 GMT http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/74913 2019-01-01T00:00:00Z http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/103862 標題: 黑箱最佳化中的隱含限制條件：雙代理模型機制; Unrevealed Constraints in Black-Box Optimization: A Bi-surrogate Scheme 作者: 湯琦恩; Chi-En Tang 摘要: 最佳化複雜且大規模的系統時，常常難以建立具備解析解的目標函數，因此需要仰賴模擬器。然而，執行高擬真度的模擬器的計算成本卻極為高昂。雖然引入代理模型來近似適應度地形，可大幅緩解此評估成本。然而，商業用高擬真度的模擬器通常會基於現實的物理或製造限制，先修復解的可行性後再評估其品質，而這些限制因受商業保護而無法取得。由於模擬器的黑箱修復具有高度的非線性，若直接對原始解進行建模，可能難以近似這種複雜的隱藏變換。這種理論上的學習瓶頸可能需要夠大的訓練資料集才能克服，反而導致評估成本增加。為克服此挑戰，本論文提出一個以雙代理模型架構為核心的最佳化框架。該框架透過結合學習修復機制與評估機制的代理模型，將修復過程與目標函數分開建模，使代理模型可以直接從修復後的可行解空間中學習。在本研究的實驗中，結果表明此方法能顯著提升學習成效，在不犧牲解品質的前提下，減少了高達 89% 的模擬器評估次數；且當模擬器的計算成本越高時，此效率增益便越顯著。; Optimizing complex systems frequently lacks analytical objective functions, necessitating simulation-based optimization. However, executing high-fidelity simulators incurs extreme computational costs. While machine learning surrogates approximate the fitness landscape to mitigate this bottleneck, commercial simulators enforce physical constraints by repairing solution feasibility prior to evaluation. Because these unrevealed mechanisms exhibit mathematical non-linearity, direct surrogate mapping of raw solutions potentially introduces learning complexity. This difficulty demands massive training datasets, resulting in an inflated computational overhead. To address this challenge, this thesis proposes a bi-surrogate framework. It deploys a repair surrogate to model feasibility and an evaluation surrogate to approximate fitness. This separation allows learning directly from the repaired feasible space. Empirical results demonstrate enhanced performance, reducing simulator evaluations by up to 89% without sacrificing solution quality. Furthermore, this efficiency gain becomes increasingly under costlier simulators. Thu, 01 Jan 2026 00:00:00 GMT http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/103862 2026-01-01T00:00:00Z