Chung Cheng University Institutional Repository
Not a member yet
889 research outputs found
Sort by
利用鋁合金冷作平面應變擠製探討摩擦界面硬化層現象;Mechanical Properties of a Harden Layer Generated in Cold Plane Strain Extrusion of Aluminum Alloy 6061
No full text[[abstract]]在金屬成形製程中由於加工時材料表面承受極高應力及應變率下，摩擦界面處通常會發生微觀結構的重排，導致在金屬表面層上產生晶粒細化與硬化層。相較於典型的塑性變形層，摩擦界面處之薄層的機械性質變化的梯度極高，因此長久以來一直是理論發展上的一個瓶頸，需要發展一套在理論及實驗上皆能應用的新方法。另一方面，機械零件的成形製程中材料的微觀結構表面層對產品品質有顯著的影響，因此發展可預測金屬成型中的摩擦界面材料特性方法是力學與機械加工的重要課題。 本研究以應變率強度因素為理論基礎，因為等效應變率梯度在靠近摩擦表面非常高與機械性質變化所展示的高梯度之實際現象亟合。由於應變率強度因素是奇異項的係數，有限元素法的套裝軟體並不適用，所以本研究團隊發展特徵值數值分析方法來計算特徵值應變率強度因素，在特徵值坐標系中進行數值分析可以避開應變率強度因素的奇異點，建立應變率強度因素的數值運算法和摩擦界面處薄剪流層內材料機械性質的本構方程式，目前可以求得平面應變下塑性加工的材料的應力應變。 此論文的目標是發展一套實驗流程以平面應變為條件進行模具設計，並選取鋁合金為加工材料，在不須施加加工溫度的條件下探討不同平面應變時所產生硬化層的關聯性以驗證應變率強度因素的數值分析方法，進而開發出能夠預測材料在密集應變後表面硬化層厚度的方法。實驗主要是透過量測材料的硬度和平均晶粒尺寸來探討整個製程設計對應變率強度因素的影響，以金相實驗來觀察成品晶粒分布情況，再藉由奈米壓痕試驗來量測邊界硬化層的厚度分布，而實驗結果將提供足夠的數據來估算應變率強度因素的大小。最後將實驗數據結合應變率與硬化層厚度的的數值計算方法來估算應變率強度因子(D)值，並利用此應變率強度因子計算出鋁合金在擠製加工後的硬化層厚度分佈，進而驗證此數值分析理論。
Intensive microstructural rearrangements usually occur near frictional interfaces in metal forming processes, it is because the surface of the material under extreme stress and strain rate. These rearrangements result in the generation of fine grain and hardened layers. One of the main contributory mechanisms responsible for the generation of these layers is severe plastic deformation. As compared to typical processes of severe plastic deformation, an exciting challenge of describing the evolution of material properties near frictional surfaces is that the gradient of these properties is extremely high at some distance from the surface. This suggests that existing approaches to study material behavior in deformation processes can be inadequate and calls for new approaches in theoretical and experimental research. On the other hand, the microstructure of material in narrow sub-surface layers has a significant effect of the performance of structures and machine parts produced by metal forming processes. Therefore, it is of great practical importance to develop an efficient approach to predict the evolution of material properties in narrow layers near frictional interfaces in metal forming processes. The theoretical basis is the strain rate intensity factor, found that the gradient of the equivalent strain rate is very high near such surfaces, which is in qualitative agreement with experimental results that demonstrate a high gradient of material properties. Since the strain rate intensity factor is the coefficient of a singular term, commercial finite element packages are not capable to provide the solution for this quantity. The numerical code to be developed is based on the method of characteristics. The expression for the strain rate intensity factor in the characteristic coordinates should allow one to eliminate the singularity in the numerical code, and our series of endeavors is the development of a numerical method for calculating the strain rate intensity factor in plane strain condition and the development of new constitutive equations for predicting the evolution of material properties within the thin shear layer or hardened layer near friction surfaces. The new constitutive equations will connect the strain rate intensity factor and parameters characterizing material properties within the fine grain or hardened layer. This dissertation develops a experimental processes to conduct cold extrusion of aluminum alloy under plane strain condition, which could deform the metal under plane strain without any temperature elevation, investigates the relationship between the harden layer produced by different strain with aluminum alloy is discussed, verifies strain rate intensity numerical analysis method, and develops a mechanistic approach to describe the mechanical behavior of the affected layer near the frictional surface. The mechanical properties were investigated with nano-indentation and metallographic experiments. Finally, the result obtained for the material coefficient α show that the theory based on the strain rate intensity factor(D) is applicable for describing the formation of the harden layer. Using this theory it is possible to determine the thickness of the harden layer in aluminum alloy extrusion processes and verified the numerical analysis theoretical
離子高分子合金複合材料製作之雙節式致動器於水中無線控制之應用研究;Application of Double-section Actuator Made of Ion Polymer Metal Composite in Water Wireless Control
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的應用，透過元件的致動，達成於水中無線控制的致動器之效果。 以離子性高分子合金複合材料製作之雙節式致動器，應用於水中探測等領域上，以元件尺寸小以及親水性特性，使得應用於此領域具有相當優勢。 為了驗證元件在水中探測的應用，在實驗水池中進行無線控制，元件游動至另一端後返回，進行致動元件在水中的來回運動。關鍵詞: IPMC致動器、雙節式、水中探測
This paper mainly studies the application of ionic polymer alloy composites, through the actuation of components, and achieves the effect of wirelessly controlled actuators in water.??? The use of ionic polymer composite materials made of double-section actuator, used in water detection and other fields, with small component size and hydrophilic properties, making the application in this area has considerable advantages.??? In order to verify the application of the components in the water detection, in the experimental pool for wireless control, the components swimming to the other end after the return of the actuator in the water back and forth movement.Keywords: IPMC actuator, double-section, water detectio
臺灣設置兩岸信心建立措施與SWOT分析研究;The Research of Cross-Strait Confidence-Building Measures and SWOT Analysis
No full text[[abstract]]1949年後，兩岸形成隔海峽而分治的現實，過去，兩岸敵對且無任何互信機制情況下曾多次爆發軍事衝突，造成雙方各有死傷。之後，隨著臺灣領導人的更迭兩岸關係也有了改變，從蔣介石總統兩岸關係「漢賊不兩立」；蔣經國總統解嚴並開放國民赴中國探親；李登輝總統廢止動員戡亂時期臨時條款及海基、海協兩會協商談判；陳水扁總統的善意溝通，存異求同政策；馬英九總統與中國展開經濟（ECFA）及各項協議談判。目前兩岸關係在整體上已趨於和緩穩定。惟兩岸仍有爆發軍事衝突的可能，鑒於此，本文參考西方既有「信心建立措施」的經驗，針對臺灣建立兩岸信心建立措施上作探討，並透過SWOT分析法，分析臺灣建立兩岸信心建立措施的優勢、劣勢、機會和威脅。
Taiwan and China have been separately ruling themselves since 1949. Without mutual-trust mechanism, Taiwan and China had several military conflicts. As administration has been changing, the Cross-Strait relations get changed as well. From years of tension under Chang Kai-shek’s regime to Chang Ching-kuo’s end martial law and opening of visit to main-land relatives, the Cross-Strait relation went d?tente. From Lee Teng-hui nullification of Temporary Provisions Effective During the Period of National Mobilization for Suppression of the Communist Rebellion and negotiation between Straits Exchange Foundation and Association for Relations Across the Taiwan Straits, Chen Shui-bian’s tolerance for difference and pursuit of concord and Ma Ying-jeou’s series of negotiations including Economic Cooperation Framework Agreement (ECFA), the Cross-Strait relations gets more and more stable by degrees. Nevertheless, the possibility of military conflict still exists. Therefore, the study is to analyze the advantage, disadvantage, opportunity and threat under “Confidence Building Measures” by SWOT Analysis
考量既定資源限制之卷積神經網路硬體設計取捨;Trade-offs and Optimization Strategies for Resource-Limited Convolutional Neural Network Hardware Design
No full text[[abstract]]近年來，由於深度學習的發展，卷積神經網路（convolutional neural network，CNN）已經廣泛地被應用在圖像辨識上。但因為本身的計算較為複雜，因此會需要硬體化來滿足某些應用上的效能需求。由於FPGA具有高性能、可重新配置、開發快速的優點，各種基於FPGA平台的硬體加速器被提了出來。但硬體加速器可能的設計方案太多，若沒有好好考量的話，加速器可能會因為沒有充分利用邏輯資源或是記憶體頻寬，而無法達到最佳性能。而且每個FPGA平台的資源限制皆不同，若要將設計好的硬體移植到別的FPGA上，可能會遇到資源不夠或是無法有效運用資源的情形。因此我們將卷積神經網路的設計利用循環分塊(loop tiling)等技術優化，並量化計算吞吐率、所需的記憶體頻寬、資源使用率。然後在Roofline模型[11]的幫助下，找到最佳性能和最低FPGA資源需求的設計方案。所以在移植到新的FPGA平台之前，可以先依照新平台的資源限制事先調整設計。若整個FPGA除了卷積神經網路外還有其他硬體的話，也能夠考量要分配多少資源給其他硬體使用，然後改變卷積神經網路的設計。
Recently, the convolutional neural network (CNN) has been widely used in deep learning for many challenging tasks, such as image recognition. Because of complicated calculations, CNN often needs to be implemented on FPGA, GPU or ASIC to meet the performance requirement. Among these realization alternatives, FPGA has been accredited for high performance, reconfigurable, and short development. Consequently, FPGA based CNN accelerators deserve good optimization strategies in order to achieve high performance under logic, memory, and I/O bandwidth constraints.In this regard, we propose to use loop tiling and subsequently to calculate the throughput, memory bandwidth, and resource usage, all under the Roofline model. As such, we can easily find trade-off among various design parameters. Moreover, the proposed methodology can be quickly adapted to other platforms for the same purpose of prototyping CNN accelerators in FPGA
不規則ROI雙鏡頭高動態範圍影像合成;Irregular ROI Dual camera HDR Synthesis
No full text[[abstract]]傳統的高動態範圍影像合成，是以一系列連續且不同曝光時間的影像來合成，以輻射照度域對應(radiance domain mapping)和曝光融合[1]兩類為主。過去方式大多是假設輸入影像為完美對齊，但實際連續拍攝場景並無法完美對齊。我們提出了針對不規則ROI區域的高動態範圍影像合成技術，以Dual camera同時設置兩張來源影像[2]，分別為自動曝光(automatic exposure)與補償曝光(complimentary exposure) 兩張不同曝光與視角的影像。過去的影像對齊方式使用尺度不變特徵轉換(Scale-invariant feature transform)，在相同曝光時間影像中找到特徵點並加以對應轉換，但在對應曝光時間差距較大影像會難以找到對應特徵點，導致對齊困難。我們提出區域亮度分部一致性來尋找對應特徵區域，解決以往全域搜尋對齊時間較長的問題，再以細部pixel劃分出ROI邊緣，準確勾勒ROI外形。在針對亮度高反差場景的情況，本篇論文題出了區域曝光補償的方法，增加ROI邊緣與背景的亮度色彩連續性，解決高動態範圍合成時ROI邊緣亮暗色彩不連續之情形，以達成高動態範圍影像合成。最後提出了在ZYNQ-FPGA板上，以linux系統為基礎，透過DMA資料搬移的軟硬體溝通環境模型，以供日後將演算法在FPGA上實現。
Traditional high dynamic range image synthesis methods are acquired from a sequence of differently exposed images. Most of them are set with static scenes and assume the all input images are perfectly aligned. However, these methods are not effective to cope with practical scenario that the misaligned situation occur.In this paper, we present an effective HDR image synthesis technique for irregular ROI form two different exposed images, one for automatic-exposed and the other is complementary-exposed. For reducing the images capture timing, we make the images captured by dual camera in the same time, so that we can make sure the scenes be static in the least capturer timing. It becomes the problems that derived from the different viewpoint and different exposure timing of two image. We propose a matching method that overcomes those problems, not only match the feature region in two misaligned images but label the ROI edge shape by pixel, which become more faithful to the over exposured region. To over come the scenes within high contrast brightness, we propose the Brightness Compensation Image Synthesis, coping with the color discontinuous and brightness discontinuous around the ROI. In the last chapter we propose the environment on ZYNQ-FPGA,useing DMA with user-space. After understanding the AXI bus in Zedboard, We can conveniently transfer data between the ARM, PL and DDR so that we can realize our algorithm in the future
以影像基礎特徵進行物件追蹤;Object Tracking Using Low Level Image Features
No full text[[abstract]]物件追蹤在電腦視覺領域是一個重要的研究，目的是在連續的影像中，不間斷地追蹤使用者有興趣的、特定的物件。本論文嘗試以影像基礎特徵進行物件追蹤的物件追蹤演算法，主要是利用針對每一個物件本身都會存在特徵點，在影像中選取要追蹤的物件之後紀錄特徵點的資訊，並將兩兩相鄰的影像中的特徵點匹配，以此不斷追蹤物件，並在被追蹤的物件離開影像範圍或是被遮住之後取消追蹤，並用記錄的特徵點資訊不斷在影像中重新搜尋欲追蹤的物件，當物件重新出現時繼續追蹤。實驗結果證明提出的演算法可以有效的即時追蹤物件，並能適應物件的變形，以及實現物件的再追蹤。
Object tracking is an important research field in computer vision in that it continuously identifies and records objects of interest in video frames. This thesis attempts to do object tracking using only low-level image features. The basic idea is to employ the low-level features to compose the so-called singular points, and to use these singular points as the basis for matching the object of interest in adjacent image frames. We demonstrate that if the tracking region surrounding the object of interest is well defined and maintained, the proposed method can provide effective tracking even under the conditions of minor obscurity, deformation, and re-entry
利用RGB-D 攝影機及離散邊緣點之視覺里程計;Sparse Edge Visual Odometry with an RGB-D Camera
No full text[[abstract]]我們提出了基於彩色與深度圖像（RGB- Method）將特徵匹配與相機運動計算合併進行。我們可以根據光度誤差（Photometric Error）調整光束平差法(Bundle Adjustment)權重以減小離群點產生的影響，然而每個特徵點都會受傳感器的深度噪聲影響，使計算相機姿態時會產生誤差，所以我們使用對極幾何模型與混合分佈模型來更這些特徵點的深度值。根據深度值的更新結果，我們可以使用N點透視法（Perspective-n-Point, PnP） 再次調整相機姿態，減少相機運動在實際運動軌跡上的漂移。我們最後以實驗結果評估本篇方法在修正軌跡的性能與可行性。關鍵詞：視覺里程計、直接法、關鍵幀
We propose a sparse edge visual odometer model based on the RGB-D camera using direct method to match intenity of the image. Compared with the we use the edge information on the image to keep the robustness of the matching in environment, and merging feature matching into pose estimation. According to photometric error, we can revise the bundle adjustment of weight to reduce the effect of outliers. However, all points is affected by the depth of the sensor noise, that are still producing difference for motion estimations. So we use epipolar geometry and mixture model to update. We will use PnP to adjust pose according to result, and reduce camera motion drift. Finally, evaluate our method in the experiment.Keywords: Visual Odometry, Direct method, Keyfram
Broadcast Markup Language 於嵌入式系統之實作與應用;Implementation and Application of BML in an Embedded System
No full text[[abstract]]科技的進步使電視的發展有了很大的改變，從原本的有線傳輸、類比訊號發展到如今的無線傳輸與數位訊號，隨著數位機上盒(Set-Top Box，STB)的普及，無線數位電視成了民眾免費訊息的來源，無線數位廣播的優點可以傳輸各種不同的資訊、不會受時間與空間的限制，且可以在不同平台上使用，因此除了單純的電視頻道外，無線數位廣播也可以加入互動式的內容，像是天氣內容的查詢、政府相關法規的推播等等，都可以透過無線數位廣播傳遞給民眾。本論文使用揚興科技提供的ISB293 STB開發板做為嵌入式系統的開發平台，藉以實現播放電視節目廣播及瀏覽BML資訊的功能。首先針對系統開發套件進行重新組譯的動作，之後將編譯好的系統以USB或USB ICE Tool燒錄到平台上完成系統的建置。平台建置完畢後，首要任務是將BML browser SDK移植到平台系統上，完成移植後需要把平台底層相對應的API與SDK作連結，若無API與SDK沒有相對應的功能則需要調整或是開發，完成上述步驟後得以實現播放無線數位電視頻道與BML的功能。在平台上完成BML功能後，由於SDK在移植與系統速度上，有一些改善空間。因此藉由BML browser的開發嘗試找出在速度上可以優化的地方，且在BML browser完成後，後續在移植到其他平台時可以節省較多開發時間。
With the progress of technology, the development of television has been greatly changed from the original cable transmission, analog signal to today's wireless transmission and digital signals. By the popularity of Set-Top box, the Digital TV has become the source of free messages for people. The advantages of digital broadcasting can transmit a variety of different information and will not be limited by time and space. It can also be used on different platforms. In addition to simple TV channels, the digital broadcasting can even add interactive content, such as the weather and important laws. Each kinds of the messages can be transmitted through the digital broadcast to the public.In this paper we use the ISB293 STB development board provided by Aonvision Technology Corporation as the development platform of embedded system. In order to realize the function of broadcasting TV and browsing BML information. First of all, we recompile the system development kit, then upgrade the system by using USB or USB ICE Tool. After building up the system of the platform, we need to add BML browser SDK to the platform system by linking the underlying API. To complete the above steps we can achieve the ability to play digital TV channels and BML.BML function which complete in the platform has improvement of the SDK transplant and system speed. So by BML browser development, we attempt to find out where the speed can be optimized. And the completion BML browser can save more development time in the transplant to other platforms
考慮複循環機組等效運轉時數(EOH)之短期機組運轉排程研究;Short-term Unit Commitment Considering Equivalent Operation Hour (EOH) of Combined-Cycle Units
No full text[[abstract]]近年來，國際能源價格成本攀升，促使發電業者不得不朝向提升營運效率和降低電能生產成本的方向邁進。然而，因為複循環技術的實現，大幅提高燃氣發電的熱機效率，使得複循環機組被大量建置，但是複循環機組同時也帶來運轉排程上的不確定性，進而影響機組排程規劃。本文建立一個考慮機組運轉特性的機組排程模型，以最小發電成本為目標函數，然後以PYOMO工具包建構數學模型，使用分段線性化得到近似機組輸入和輸出的線性直線，並使用混合整數線性法規劃線性模型，再輸入至解法器GUROBI求解線性系統模型。
In recent years, power industries have been aggressive in reducing costs or improving operation efficiency because of higher international energy's prices. Fortunately, gas power generation benefit from the realization of the combined cycle technology, which greatly improve the power generation efficiency. However, it will become more difficult to do electrical dispatch when a large of combined cycle units are built.In this thesis, Pyomo is adopted to establish unit commitment model with short-term scheduling of combined cycle plants and thermal units by considering Minimum cost and generator characteristics including combined cycle unit’s Equivalent Operation Hour. The mathematical model proposes Mixed Integer Linear Programing (MILP) for based approaches, and piecewise linear approximation is used to get units input/output curves
應用於人體動作偵測與分析之都卜勒雷達設計;The Doppler Radar System Designed for Human Motion Detection and Analysis
No full text[[abstract]]本論文乃開發一步態偵測之微波生醫雷達. 主要針對實驗室現有的兩套雷達系統設計訊號處理演算法，分別為一維四波束雷達系統以及二維波束雷達系統，驗證其空間切割之能力。而於一維四波束雷達系統上首先遇到的問題為無法將左右相鄰波束訊號做區分，但若是利用濾波器濾除低頻身體訊號後，即可發現左右波束接收到的手部訊號依然擁有強度上之差別，而藉由左右波束之頻率成分強度不同，將左右波束接收到的訊號區分開，接著將一維四波束雷達系統的演算法經驗代入二維波束雷達系統，發現可以順利區分手部之訊號但腿部訊號因為雙腿距離過近導致波束接收訊號重疊現象產生，故利用手腳關聯性來將雙腿訊號區分開來。實驗中僅使用了二維4×4總共十六個波束中的四個波束，以驗證其空間切割之能力，確認了中間四波束正常使用之後可以更進一步朝周圍的十二個波束發展，使得二維波束雷達系統更為全面可以偵測更多的人體動作。
This thesis focuses on the development of microwave bio-radar for gait detection. The signal processing algorithms are newly proposed for spatial separation. The problem first encountered is that the one-dimension four-beam radar system can’t separate signals from two adjacent channels. After using a digital filter to filter out low frequency body movements, the signal between adjacent channels still have some difference in magnitude. By examining the magnitude from different frequency contents, we can successfully separate the signals of left arm and right arm.The system is further extended from one-dimension four-beam radar system to two-dimension sixteen-beam radar system, aiming to capture the signals from more body parts. Although the signals from both left arm and right arm can be observed, the signals from legs cannot be distinguished due to larger RCSs. The correlation of arms and legs is therefore applied to solve this problem.This research only use the middle 2×2 four beams out from 4×4 sixteen beams to verify the ability of spatial separation. based on this outcome, we can enhance the spatial beam resolutions to detect more details of human gait motion