Chung Cheng University Institutional Repository
Not a member yet
889 research outputs found
Sort by
論我國商標法對著名地方特色產業之保護;
No full text[[abstract]]我國歷史悠久且具有豐富多樣的自然環境，符合塑造地方特色產業之條件，除經歷長久時間發展而來者外，自民國78年起，我國即開始進行「一鄉鎮一特產」運動，因此，形成許多在地歷史性、文化性、獨特性或唯一性與依賴當地特殊自然環境生成等特質的地方產業。根據經濟部統計，我國著名產業產地共有346項，這些地方特色產業之存在反映了當地特殊歷史、文化、風俗、習慣或氣候、地形、土壤、海拔等人文背景或自然環境，其不但具有傳承地方文化或技藝之功能外，也是地方經濟及社會穩定的支柱，並在全球貿易自由化下，能提供在地就業機會，使青年人口回流，繁榮當地經濟，達到挑戰全球市場之終極目標，成為國家經濟之後盾，是重振地方經濟不可或缺之角色。故，對我國而言，地方特色產業相當重要，有提供其保護之必要。由於地理標示與地方特色產業之特性共通，本文以地理標示之角度探討產地及地方特色產業之保護。目前我國地方特色產業，包含地方特色產品及地方特色服務業，獲得保障之主要途徑為註冊產地證明標章及產地團體商標。然而，並非所有地方特色產業都能達到受產地標章保護之水準，例如花蓮麻糬、社頭襪子及五股家具等，且此類地方特色產業多為由人文因素所形成者，但有鑑於地方特色產業之經濟及文化價值，對於無法受到產地標章保護之地方特色產業，本文認為仍應提供其積極保障。日本傳產法以「於一定地域形成產地」為條件，透過「傳統證紙」商標保護地方特色工藝品；歐盟傳統特產保證制度則明確訂定符合傳統特產之要件者，專有名稱使用權及排除名稱不當使用權，保護傳統農特產及食品，兩者於實行上效果顯著。故，於參考上述制度後，本文認為我國應以專法建立類似制度，明訂「於一定地域形成產地」或「在特定地域為服務提供地」為保護要件，區分農產品及食品、地方特色工藝品與服務三種不同之保護客體，分別提供類似於傳產法或傳統特產保證制度之保護，使未能受產地標章保護之地方特色產業能直接獲得保障。
Taiwan has a long history, rich and varied natural environment, which meets the conditions of shaping the local special industries. With developing for a long time and the government carrying out "One Town One Product " movement since 1989, there are at least 346 well-known local industries in our country. They can not only inherit the local culture or traditional skills but also help stabilize the local economy and society. In short, they play an indispensable role to revive and prosper the local economy. Hence, it is necessary to provide them with protection. Due to the common characteristics of geographical indications (GIs) and local special industries, the author dicusses the protection of the origins and the local special industries in the view of GIs. Currently, the main approach to local special industries (including products and service) protection in our country is to register the geographical certification/collective marks. However, not all of them can meet the requirements of the geographical marks, such as Hualien Mochi, Shetou hosiery and Wugu furniture, etc. Mostly, such local special industries are formed by the human factors. In the light of the economic and cultural values of local special industries, the author considers that it is still necessary to provide positive protection for these kinds of local special industries.Japan’s Densan Act protects local special crafts through Densan mark, which aquires the crafts to meet the condition of "becoming origin in a certain region". The EU’s TSG system protects the traditional specialities which include agricultural products and foodstuffs. As long as a product meets the requirements of the traditional specialty, it will be provided with the exclusive right to use the name and the right against any practice liable to mislead the consumer. Both systems have made great success. Therefore, the author concludes that our country should establish a similar legal system, stipulating the requirement of "becoming origin in a certain region" or "providing service in a certain region" so that the local special agricultural products, foodstuffs, crafts and service which can not meet the requirements of the geographical marks will be provided with a direct protection
商品責任之比較研究 - 以越南和台灣法為中心;A Comparative Study of Product Liability Between Vietnam and Taiwan
No full text
加權圖中Closeness Centrality前k高之估算方法;Estimating Nodes of Top-k Closeness Centrality in Edge-Weighted Graphs
No full text[[abstract]]近幾年來，在各種網路圖上作centrality的分析是討論度極高的議題，因為centrality的數據能在各種應用中都能提供相當的幫助。而在一些centrality(例如:betweenness及closeness)的計算中都是基於最短路徑的計算，但由於現今的網路圖(尤其是社群網路的圖)的資料量都相當龐大，為了計算centrality的最短路徑計算就會變成一個相當耗時的工程，所以在很多研究中都在修改最短路徑演算法，以此希望能大幅度減少前置的最短路徑計算時間，且同時在估算前k個centrality較高的點是哪些時，又可以得到相當不錯的命中率，而大部分的研究都著重在無權重的圖上做分析，在本文中將提供一些修改的方法使得原本在無權重圖才能用到的修改方法可以使用在有權重的圖上，又能得到不錯的結果，且本文的研究將會著重在closeness centrality上。
Social network analysis is a hot issue in recent years. Centralities are usedto represent the importances of nodes. By definition, the closeness centralityinvolves computing the total distance to all other nodes and is therefore verytime-consuming. In this study, we focus on estimating node of top-k closenesswithout computing all-to-all distances. For each node, instead of computingthe shortest paths to all other nodes, we define the "local closeness" whichinvolves the number of nodes and their distance within a specified radius, andwe use the local closeness to estimate the nodes of top-k closeness
支援時序猜測架構之迴圈轉換與指令排程技術;Loop Transformation and Instruction Scheduling Techniques for Timing Speculative Architecture
No full text[[abstract]]傳統的處理器會嚴格地限制工作的電壓與頻率以確保在最糟糕的條件下保持處理結果的正確，但隨著製成的進步與電晶體密度的提高，為了應對環境的變動衝擊，電壓與頻率限制的越加嚴格。時序猜測架構透過允許處理器可以產生錯誤結果並透過錯誤容忍機制偵測與修正錯誤以降低電壓與頻率的限制瓶頸，然而程式會因時序錯誤而造成執行效能的影響。 本研究可分為三大工作。第一，透過實驗平台分析程式指令的使用行為對時序錯誤的影響，以進行對時序猜測架構優化。第二，透過分析結果提出迴圈轉換以有效降低高達37%時序錯誤的產生。第三，透過分析結果提出指令排程技術以分散時序錯誤的分布，以增進時序猜測架構與自適應電壓縮放技術的協作與降低高達45%時序錯誤的產生。本研究將研究成果實作於開源的LLVM編譯器基礎設施上，並能夠自動地使用我們的迴圈轉換與指令排程技術產生程式。
Traditional processor design incorporates voltage and frequency guardbands to ensure correct execution of operations under worst-case conditions. As transistor density increases and manufacturing processes improve, increasingly costly guardbands are required to deal with the impacts of environmental variability. The use of timing speculation can relax the tight constraint for worst-case design by allowing occasional errors, which are detected and corrected later by an error resilience mechanism. However, a program's performance may suffer owing to timing errors. The thesis consists of three works. First, we analyze program behaviors and observe what influence the number of timing errors through a simulator. Second, we propose a loop transformation technique for Timing Speculative Architectures to reduce up to 37% the number of timing errors. Third, we propose an instruction scheduling technique to rearrange the instructions in the programs that make better cooperation between Timing Speculative Architecture and Adaptive Voltage Scaling technique and reduce up to 45% the number of timing errors. These proposed techniques are implemented in the LLVM compiler infrastructure to generate the optimized programs automatically
基於深度學習應用於鋼胚瑕疵檢測;Steel Surface Defects Detection based on Deep Learning
No full text[[abstract]]鋼胚表面瑕疵檢測在鋼鐵製造業的品質管理扮演了舉足輕重的地位，然而利用人力檢測鋼胚瑕疵卻大幅減緩產品線之製造速度並耗費許多人力時間上的成本。目前傳統影像處理方法被提出來自動檢測熱軋鋼材表面，大致上分成兩步驟，影像前處理與分割瑕疵部分，前處理部分為克服影像明亮度不均的問題而分割部分則是利用影像處理的方法進行判斷並以二元圖表現出影像中瑕疵與否。這方法高度依賴特徵擷取的過程，但瑕疵的特徵擷取卻不容易去選取判定。在本篇論文中我們提出根據深度學習之應用來自動偵測鋼胚影像瑕疵，不像傳統方法利用人為設定瑕疵而是透過大量訓練資料來讓機器學習具代表性之特徵瑕疵，並依檢測物件不同來優化其檢測能力，而此方法也能應用於其他工業檢測上。
Surface defects detection plays a significant role in quality enhancement in steel manufacturing. However, manual inspection of steel surface slows down the entire process and time consuming. Currently, many methods had been proposed for automatic defect detection on hot-rolled steel surface. These methods usually follow two steps: pre-processing and segmentation. The pre-processing step is to overcome the uneven illumination of images and the segmentation step is to generate the binary map to identify defect. This kind of methods highly depend on feature selection approaches, but the defect features are usually not easy to obtain. In this thesis, we propose an automatic steel surface defects detection method based on deep learning to replace traditional way to image processing. The deep learning models will be evaluated for defect detection. The experimental results show that the proposed method can detect steel surface defects more effectively and accurately than the traditional methods, and this approach can be also applied to other industrial applications
基於深度感知模型的遠紅外光人臉影像辨識之參數研究;A Parametric Study of Deep Perceptual Model on Visible to Thermal Face Recognition
No full text[[abstract]]隨著影像技術的蓬勃發展，多年前人臉辨識系統就能達到近乎百分之百的準度，但人臉辨識的效能常受光線變化的影響，在光線不足的狀況下，人臉辨識變得更加困難。正因如此，近年有許多關於人臉辨識的研究使用了紅外線的人臉影像來實踐。在本篇論文中，我們選擇使用可見光及長波長紅外光影像 (LWIR) 來做人臉匹配，由於長波長紅外光影像是靠熱能散發來成像的，所以就算在沒有光線的場景中，也能靠溫度產生出影像，此種特性特別適合應用在夜晚拍攝的監控系統上。 近年來發表的長波長紅外光人臉辨識研究中，基於自動編碼器的深度感知映射模型 (DPM) 達到了最先進的效能。他們從長波長紅外光人臉影像中的區塊提取特徵，並經由自動編碼器將其轉換至另一個向量空間中，使得轉換後的紅外光影像與可見光影像的特徵更相似。在本篇論文中，我們評估了不同參數的DPM，以便為將來的研究建立實驗細節上的參考。 在實驗評測中，我們驗證了模型應用在不同測試資料上也能夠維持辨識的有效性，並且證明將兩個不同型態的影像對齊，可以進一步提升紅外光人臉辨識的效果。最後，也探討本系統的參數設定對於預測效果的影響。
Owing to the advance of image technology, face recognition has achieved high performance years ago. Face recognition is highly influenced by illumination, and it becomes more challenging when the face image is captured in the dark or in poor light. In order to avoid the influence of illumination, recently infra-red images are used in many face recognition studies. In this work, we take long-wave infra-red (LWIR) as our target. Infra-red images are imaged based on temperature, so we can capture infra-red image even in the dark. With this property, it may be applied in a night-time surveillance system.Recently deep perceptual mapping (DPM) based on deep neural network provides the state-the-art thermal to visible face recognition. Features extracted from patches of a long-wave infra-red face image are transformed into a vector space by a deep neural network, such that features from infra-red images are comparable with features from visible images. In this work, we comprehensively evaluate DPM with different settings, in order to build a reference study for future research.In the evaluation, we verify the model does have the ability to adapt to different data, and investigate the influence of model parameters on prediction performance
離子高分子金屬複合材料蠕動致動器設計製造及測試;
No full text[[abstract]]本篇論文主要在研究離子性高分子合金複合材料(Ionic Polymer Metal Composites,IPMC)的製程與應用，透過元件與各種工具組裝後，形成類似仿生食道。在製程方面，文中對於IPMC致動器的整個製程、與各個工具製程做了完整的介紹。在本篇論文中，提出了一種多段式可彎曲的致動器，元件本身採用高分子金屬複合材料(IPMC)，在組上我們新開發的工具，來產生類似食道蠕動的功能，在通過水熱法的加工使其與傳統的IPMC相比有著更好的韌性，結合已3D印表機印出的自製扣環，與二甲苯薄膜電線的結合，能更方便的給予電極，將丙烯?胺水凝膠管與IPMC致動器一起作為仿生食道模擬進行測試，通過IPMC三節的相位控制，來達到食道蠕動的特性。
This thesis mainly studies the process and application of Ionic Polymer Metal Composites (IPMC), which is similar to the biomimetic esophagus after assembly with various tools. In the process, the text for the IPMC actuator of the entire process, with the various tooling process to do a complete introduction.In this paper, we propose a multi-stage flexible actuator, the element itself using polymer metal composite (IPMC), in the group of our newly developed tools to produce similar function of esophageal peristalsis, through the water thermal processing to make it with the traditional IPMC has a better toughness, combined with a 3D printer printed self-made buckle, and xylene film wire combination, can more easily give the electrode, the acrylamide water The gel tube was tested along with the IPMC actuator as a bionic esophageal simulation, and the phase control of the IPMC three sections was used to achieve the characteristics of esophageal peristalsis
以即時穿透式電子顯微鏡技術量測類鑽碳吸附力的研究;Research on measurement Diamond-Like-Carbon by In-situ Transmission Electron Microscopy
No full text[[abstract]]類鑽碳鍍層是一種同時具有低摩擦力和低磨耗之保護材料的重要鍍層，廣泛的應用於如硬碟和刀具以及引擎元件等領域。吸附力的量測是探討磨潤現象的主要基礎，以往量測吸附力最重要儀器為原子力顯微鏡(AFM)，但是原子力顯微鏡在大氣環境下樣品容易吸附汙染物，而且不能精確控制接觸點和時間。因此本研究應用穿透式電子顯微鏡來進行即時奈米定位量測，測量類鑽碳的吸附力。 本研究在穿透式電子顯微鏡(TEM)中進行即時量測，測量鑽石探針與原子力顯微鏡探針上鍍類鑽碳薄膜在接觸和脫離時的瞬間吸附力，即兩探針接近時的拉進力(Pull-in force)與脫離時的拉脫力(Pull-off force)。透過奈米壓痕儀控制壓電管移動鑽石探針與原子力顯微鏡探針接觸，但奈米壓痕儀中的力量感測器精度不夠不足以測量吸附力，因此應用精密定位及即時量測來測量鍍有類鑽碳之懸臂梁探針在接觸和脫離時瞬間彎曲的角度，此角度與臂梁探針的彈簧常數乘積即為拉進力與拉脫力。實驗中探討推進距離與停滯時間對吸附力的影響，結果顯示各種實驗情況下的拉進力值皆非常相近不受粗糙度影響。由實驗結果推導出拉進力的哈梅克常數(Hamaker constant)，與理論上鑽石對鑽石的哈梅克常數十分接近，由此可以推論拉進力主要為凡得瓦爾力(van der Waals’ force)。在脫離時的拉脫力在不同推進距離與停滯時間的值顯示較為分散，由量測數據推論是受到類鑽碳表面粗糙度影響，粗糙度越大拉脫力越小。 本研究接著使用分子動力學模擬(Molecular Dynamics Simulation)進一步驗證實驗結果，模擬中探討論了類鑽碳表面粗糙度對拉進力與拉脫力的影響。模擬?果顯示拉進力值在不同表面粗糙度條件下變化不大，與實驗時拉進力保持不變的趨勢一致。而且分?從模擬和實驗結果計算的哈梅克常數與理論上鑽石對鑽石的哈梅克常數接近，證明了拉進力為凡得瓦爾力。然而模擬?果顯示拉脫力的值與粗糙度有關，且與實驗?果一致，隨著粗糙度的增加先?小後不變。此外從分子模擬?果確定了粗糙度對拉脫力的影響機制：粗糙度主要通過影響接觸面形成的原子鍵?數影響拉脫力。在粗糙度較小時，接觸面形成的鍵?數較多，接觸面分離時破壞的鍵?數較多，拉脫力則越大；當粗糙度增加時，接觸面形成的鍵?數逐漸?小最後保持穩定，拉脫力則隨之減小最後趨於穩定。
Diamond-like carbon (DLC) coating is an important coating that has both low friction and low wear which can protect material, widely used in such as hard disks and knives and engine parts. The measurement of the adhesion force is the main basis for exploring the tribology phenomenon, the most important instrument for measuring the adhesion force was atomic force microscopy (AFM) but, atomic force microscopy in the atmosphere of the sample is easy to adsorb contaminants and can not precisely control the contact point and time. Therefore, in this study, the use of in-situ the transmission electron microscope (TEM), to carry out real time Nano-positioning measurement, measurement of diamond-like carbon adhesion force. In this study, real-time measurement was performed in TEM, measurement of DLC-coats AFM tip to diamond indenter of pull-in and pull-out instantaneous adhesion force, that two tips closed call pull-in force and two tips separated call pull-off force. Through the nanaindenter to control the piezo tube that enables the precise motion of the diamond indenter to contact with the atonic force microscope tip, but the force sensor in the nanoindenter is not accurate enough to measure the adhesion force. Thereore application of precision positioning and real-time measurement can be measured with a diamond-like carbon cantilever probe in contact and separate when the moment of bending, this angle multiply with the cantilever probe spring constant is pull-in force and pull-off force. At experiment discussed push distance and holding time effect to adhesion force, the result show pull-in force values are consistent and did not affect with roughness at any experiment situation. The experiment results in derivation the Hamaker constant from pull-in force, which is very similar to that measured for the Hamaker constant for Diamond on Diamond in theoretical, the value for DLC on Diamond is expected to be similar and therefore we conclude that the pull-in force observed is the result of van der Waals interactions. The pull-off force at different push distance and holding time are highly scattered, the pull-off force effect of DLC surface roughness, the increase of roughness decreases pull-off force. This study then uses molecular dynamics simulations to further validate the experiment results, DLC surface roughness discussed effecting of the pull-in force and pull-off force in the simulation. The simulation result show that the pull-in force remains unchanged at different roughness conditions, which is consistent with the tendency of the pull-in force in the experiment. Moreover, respectively the Hamaker constant calculated from the simulations and the experiment results are closed to the theoretical Hamaker constant of diamond, proving the pull-in force is Van derWaals force. However, the simulation results show the pull-off force is related to the roughness, and is consistent with experimental results. The roughness decreases first and then decreases. In addition, the effect of roughness of the pull-off force is determined form the molecular simulation results. Roughness mainly affects the pull-off force by affecting the number of bonds formed on the contact surface, roughness mainly affects the pull-off force by affecting the number of bonds forms on the contact surface. When the roughness is low the number of bonds formed on the contact surface becomed large, the number of the bonds broken when the contact surface separated is large, and the pull-off force will be larger; As the roughness increases, the number of bonds formed on the contact surface gradually decreases, eventually stabilizing, pulling force decreases, and eventually stabilizes
基於等效輪廓誤差之雙軸線上學習控制與學習控制之轉換時機探討;
No full text[[abstract]]智能化已是近年來工具機技術發展的主要方向之一，不僅是傳統一味的大量生產，現在更注重的是精度優化及客製化。透過即時監控，大數據分析及工廠整合設計，使生產過程更為有效，使生產結果滿足各種不同需求。線上學習控制是在離線學習控制的基礎上，去發展更為節省時間的演算法。此學習控制的原理是利用前一次加工的誤差，根據與路徑相關之演算法，來修正下一次的命令，可使機台達到更高的精度；且本演算法是在不改變回授控制的前提下，於閉迴路系統外加入學習控制的功能，因此能應用在各個機台上，對機台做優化。 傳統學習控制是以追蹤誤差為目標，然而實際上能真正影響工件精度的是輪廓誤差。目前只有少數以輪廓誤差為目標的學習控制，原因在於輪廓誤差的計算並不易。本論文所提之等效輪廓誤差是以路徑方程式來建立，不僅較易記算，也可以相當程度來等效於實際輪廓誤差。 本研究所提之線上學習控制，在應用上可用於各種機台的優化，例如多軸工具機甚至機器人等。我們實際應用到了五軸工具機上，透過實驗結果可以證實，線上學習控制不僅可達到離線學習控制相去不遠的結果，在計算時間上更能縮短不少。 本研究也提出另一個優化學習控制的方向，也就是轉換時機的決定。我們透過實驗發現的最佳轉換時機會根據不同情況及條件有所變化，如果能找出其關係式，就能使學習控制在各種路徑達到最佳的效果。
Intelligence is one of the main developing direction in recent years. Not only to produce massive product, it’s more important to optimize the accuracy and customize. By immediately monitoring、big data analysis and design of the factory, we can make the producing process more efficient and the result to satisfy different kind of demand. Online learning control is based on offline learning control. The purpose is to develop an algorithm that can save more time. The principle of this learning control is to use the error of the previous processing, according to the algorithm related to the path, to correct the next command and the machine can achieve higher precision. The algorithm does not change the feedback control. Under the premise, the function of learning control is added outside the closed loop system, so it can be applied to each machine to optimize the machine. Traditional learning control is aimed at tracking error, but in reality it is the contour error that really affects the accuracy of the workpiece. At present, there are only a few learning controls that aim at contour errors, because the calculation of contour errors is not easy. The equivalent contour error proposed in this paper is established by the path equation, which is not only easy to calculate, but also equivalent to the actual contour error. The online learning control proposed by this research can be applied to the optimization of various machines, such as multi-axis machine tools and even robots. We actually apply it to the five-axis machine tool. Through the experimental results, we can confirm that the online learning control can not only achieve the results really close to that of offline learning control, but also shorten the calculation time. This study also proposes another direction to optimize learning control, which is the decision to switch timing. The best switch timing we find through experiments vary according to different situations and conditions. If we can find out the relationship, we can make learning control achieve the best results in various paths