Chung Cheng University Institutional Repository
Not a member yet
889 research outputs found
Sort by
改良式人工智慧方法於電力系統之預測應用;Improved Artificial Intelligence-Based Methods for Power System Forecasting Applications
No full text[[abstract]]近年來環保意識逐漸提升，許多國家已逐漸意識到開發運用再生能源之必要性，目前已有越來越多的國家積極投入再生能源的開發工作，其中又以取之不盡的風力能源、太陽能源之相關研究最受重視。由於隨著風力發電與太陽光電系統的建置與容量逐漸擴大，再生能源預測技術對於減輕太陽光電與風力發電之輸出的隨機性影響具有相當重要的意義，準確的再生能源發電預測則可提高電網穩定性、降低備載容量、以及電力系統運轉成本及減少污染排放，本論文針對風力發電廠與太陽光電廠之發電輸出值進行不同資料解析度之預測。然而於大型負載設備（例如：風力發電機組或電弧爐）運轉後所產生有關電力品質之電壓閃爍問題也值得探討，若這些不良的電壓閃爍嚴重程度也可以被預測，則電力公司與大型負載機組設備用戶之間可以共同利用改善設備（例如：靜態虛功補償器或熔爐控制）進行改善電壓閃爍問題。因此，本論文採用煉鋼廠的電弧爐設備為研究對象預測三種電壓閃爍指標（ΔV10, Pst, and Plt）之嚴重程度。本論文即依據上述之問題，將提出改良式輻狀基底類神經網絡（IRBFNN）、具有誤差回授機制之輻狀基底類神經網絡（RBFNN-EF）、具有誤差回授機制之改良式輻狀基底類神經網絡（IRBFNN-EF）、高斯混合模型之類神經網絡（GMMNN）、結合灰色理論之深度學習類神經網絡（Grey-DLNN）、結合灰色理論之輻狀基底類神經網絡（Grey-RBFNN）、結合灰色理論之改良式輻狀基底類神經網絡（Grey-IRBFNN）與結合查表法之灰色模型（Grey-LUT）進行風速、風力、太陽光以及電壓閃爍嚴重程度之預測應用並與過去傳統之類神經網絡方法進行比對，測試結果顯示，本論文所應用之預測方法於實際電力系統之可?性。
Increasing renewable energy penetration has been a global trend in the past decades. Wind and solar power are expected to contribute significantly to the renewable energy targets owing to advancements in renewable energy technologies, abundance of free resource and commercial viability. Therefore, the predictability of wind and solar power in managing load and generation balance is crucial to system operations. This thesis proposes improved models for wind and solar power output at different interval forecast. Also, due to the operation of the large electric devices (e.g. wind power generator, electric arc furnace, etc.) which would make the power system produce serious voltage flicker. If the flicker levels are predictable, corrective solution such as static var compensation may be developed for both electric utilities and the customer. This thesis adopts electric arc furnace at steel industrial company as investigated target for three indices of flicker severity (ΔV10, Pst, and Plt) forecast.This thesis proposes improved artificial intelligence methods to solve above-mentioned problems which include the improved radial basis function neural network (IRBFNN), radial basis function neural network with an error feedback (RBFNN-EF), improved radial basis function neural network with an error feedback (IRBFNN-EF), Gaussian mixture model neural network (GMMNN), radial basis function neural network integrated with grey theory (Grey-RBFNN), improved radial basis function neural network integrated with grey theory (Grey-IRBFNN), deep learning neural networks integrated with grey theory (Grey-DLNN), and Grey theory with look-up table (Grey-LUT). Performance comparisons between the proposed and traditional methods are reported for wind speed at 10-minute interval, wind power at 10-minute interval, wind power output at one-minute interval, solar power generation at one-minute interval, and three kinds of flicker severity level forecast. Simulated results (i.e. wind speed and power forecast, solar power forecast, and flicker severity forecast) can provide more accurate and effective forecast than other compared methods to the actual power system forecasting problems
應用於醫學超音波影像系統之寬頻連續時間三角積分調變器之架構設計與分析;Design and Analysis Wideband Continuous-Time ΔΣ Modulators for Medical Ultrasound Imaging Systems
No full text[[abstract]]本論文設計超音波微換能器與連續時間三角積分調變器應用於高解析度的超音波影像系統。超音波系統包含超音波換能器、低雜訊放大器、濾波器、類比數位轉換器以及影像處理。超音波微換能器提出五種架構分別提升最高5.8倍的輸出以及更高的線性度。連續時間三角積分調變器分別為改善額外延遲與時脈抖動提出三種新架構。預測路徑架構將量化器的預測值送回迴路濾波器，能提供接近2倍時脈週期的額外延遲容忍時間，並且達到 SNDR 70dB。改良式推疊架構利用將減少主路徑上的額外延遲又保持迴路濾波器的積分能力，並設計零點補償改善諧振器的補償係數衰減的缺點達到SNDR 83.6dB。數位積分器回授架構利用將DAC回授改在第一級積分器輸出端，使得回授路徑也獲得一階雜訊移頻的好處，可以容忍77ps的DAC時脈抖動且保持SNDR 72dB。連續時間三角積分調變器頻寬達到44kHz到8 MHz，品質因數達到0.056 J/conv.，使用180奈米CMOS製程。
This research designs the capacitive micromachined ultrasonic transducers (CMUTs) and continuous-time sigma-delta modulators, applied to high-resolution ultrasound imaging systems. The ultrasound systems includes ultrasound transducers, low-noise amplifiers, filters, analog-to-digital converters, and image processing units. Five CMUT architectures are proposed to enhance the sensitivity by 5.8 times and higher linearity is achieved. For continuous-time sigma-delta modulators three new architectures are proposed to improve clock jitter and excess loop delay. Forecast architecture is designed to send the predictive values from the quantizer back into the loop filter. Nearly double time of the clock cycle for tolerating excess loop delay is achieved, and SNDR is 70 dB. Modified stack architecture utilizes direct connection of the first stage output to the third stage integrator and provides a compensation path to the second stage output, thereby reduces the extra delay of the primary path and keep the integration capability of the loop filter, and SNDR is 83.6 dB. Furthermore, zero compensation is designed to improve the attenuation coefficients of the resonator. Digital integrator feedback architecture utilizes the DAC feedback at the first stage integrator output. The feedback path benefits from the first-order noise shaping, which can tolerate 77 ps DAC clock jitter while maintaining SNDR of 72 dB. The continuous-time delta-sigma modulator achieves a bandwidth of 44 kHz to 8 MHz with a quality factor of 0.056 J/conv.. The chip is fabricated in 180 nm standard CMOS process
應用於室內定位之非監督式無線訊號指紋圖資建立方法;Unsupervised radio map learning for indoor localization
No full text[[abstract]]過去幾十年大多數的室內定位方法多以基於Wi-Fi無線訊號指紋(Fingerprint)的定位技術為主，因為這項技術擁有低設備成本以及有效處理遮蔽問題的能力，基於無線訊號指紋的定位方法主要分成兩個階段，訓練階段與定位階段。訓練階段的目的為建立無線訊號指紋圖資(Radio Map)，建立的方法是在已知的參考位置上收集 Wi-Fi 訊號強度；定位階段則是利用當下裝置接收到的Wi-Fi強度與訓練階段的訊號指紋圖資做相似性量測得到定位結果。但若要在大範圍的室內環境提供定位服務，將需要耗費相當多的時間與人力去收集訊號強度建立訊號指紋資料庫，為了克服此問題，本論文提出一個應用於室內定位之非監督式的訊號指紋資料庫建立方法，有別於其他方法利用模擬的指紋圖資或事先取得訊號傳播模型的方式來減少花費過多人力與時間的問題，我們所提出的方法利用Wi-Fi訊號以及慣性元件訊號來幫助自動建立無線訊號指紋圖資，將慣性元件訊號處理後得到多名使用者的行走軌跡資訊與當下收集的無線訊號強度，針對不同使用者提供的資訊，利用無線訊號標記初步串接所有軌跡，並且結合四項限制來建立最佳無線訊號指紋圖資，包含軌跡無線訊號強度岐管對齊限制(manifold-based smooth)、無線訊號標記對齊限制(landmark alignment)、軌跡間相互定位限制(inter-trajectory)與位移限制(displacement)，由四項限制設計一個建立無線訊號指紋圖資的迭代優化程序，藉此程序建立完整的無線訊號指紋圖資。
For radio-based indoor localization, the approaches founded on the radio fingerprint concept are efficient duo to low cost and the ability to handle occlusion effects. However, the approaches require a lot of human labor to label training data for radio map (fingerprint) construction. To address this issue, in this paper, we proposed an unsupervised framework to learn a Wi-Fi radio map in an indoor environment. Unlike conventional approaches that depend on a simulated radio map or a prior radio propagation model to reduce human efforts, our method uses Wi-Fi and IMU signals collecting by crowdsourcing to build a robust radio map automatically. More concretely, four types of constraints are fused by the proposed radio map optimization procedure. They include the alignment of Wi-Fi landmarks, the displacement constraint, the manifold-based smooth constraint, and the inter-trajectory constraints. Our experiment results also show the effectiveness of the unsupervised radio map
基於三維幾何空間提升對應點關係的運動恢復結構;Correspondence Refinement based on 3D Geometry for Structure from Motion
No full text[[abstract]]藉由二維影像所提供的資訊重建出物體的三維架構一直為值得探討的議題，從古至今所提出的著名方法如運動恢復結構(Structure from motion)、明暗恢復形狀(Shape from shading)…等。傳統的運動恢復結構方法是將攝影機從不同位置拍攝物體所得到的影像中，找出物體特徵點的連續對應關係，從而推算出攝影機姿態，再回推出特徵點於三維空間中之位置。運動恢復結構方法是由影像間的特徵點對應關係重建出三維物體，因此當重建三維物體時，多張影像間二維特徵點對應關係所產生之特徵點軌跡(feature track)是十分重要的。但在傳統的運動恢復結構中，並無特別強調如何串聯多張影像間特徵點的對應關係，常見的方法是以共同影像間的特徵點作為連接的橋樑。這邊我們發現到若當多張影像間的特徵點對應關係中有一小段錯誤的匹配發生卻無被移除時，則此錯誤對應關係會傳遞下去，導致使用此錯誤對應關係估算出錯誤的相機姿態和三維點位置，我們稱此錯誤傳遞關係為連接錯誤傳遞(linking error propagation)。因此我們提出本論文的方法來做改善。本篇論文提出基於三維幾何空間以提升對應點關係並改善當前之運動恢復結構演算法。我們使用三維幾何空間點與點的關係來判斷對應關係間是否可歸類於同條特徵點軌跡，而非單純依影像上特徵點位置來作連接。所以剛開始假設所有影像中的對應關係均為未知狀態，接著使用傳統的運動恢復結構求得兩兩影像間的特徵點對應關係、相機姿態與三維點位置結構，此資訊我們稱為場景圖(scene graph)。接著我們對齊場景圖中提供的三維點結構，若有三維點互相十分接近彼此，我們則可確定這些三維點實際代表物體上同一位置，以此來確定所對應到的特徵點可串聯成同一條特徵點軌跡，同時可減少連接錯誤傳遞的狀況，並提升相機姿態的估測正確性。本論文最後則將所提方法之實驗結果與傳統方法進行比較，以探討其結果差異並提出未來之展望。
Reconstruct object’s 3D model from the images which are captured by the camera has been discussing for a long time, there have some well-known methods like Structure from Motion, Shape from Shading, …etc.. In the thesis, we focus on the Structure from Motion method. In this method, we have to capture the object’s images from the different viewpoint, then find the feature points’ correspondence over these multi images, based on the correspondence we estimate the camera poses, when we know the camera poses and feature points’ correspondence, finally we can reconstruct the object’s structure.From the Structure from Motion reconstruct steps, we realize that the correspondence of feature points over multi images, which is called “feature tracks”, is very important. In the traditional Structure from Motion, the general way to link the correspondence is to find the same point location at the common image. However we find out that it could occur some problems, when finding the feature track over multi images, if there has a short part of correspondence is wrong but the system treat it as inlier, which means it can not remove properly, so the wrong correspondence will propagate its error, causing the linking error propagation.We propose another way to avoid the linking errors, increase the accuracy of feature tracks, hence estimate robust camera pose. Instead of using the 2D feature points locations to generate the feature tracks over multi images, we consider the 3D-points’ geometry relationship to re-generate the 2D feature points correspondence over multi images. Based on 3D geometry, we are looking forward to avoiding linking errors propagation, so that we can estimate more robust camera poses and 3D-points locations. After that, we will compare the result of our methods with traditional Structure from Motion.In the end, we will summary our methods and make a conclusion
Affordance於機械手臂抓取之研究;A Robot Grasping Method with Affordance
No full text[[abstract]]機械手臂的控制流程，從環境偵測，物體辨識，路徑規劃，到任務執行，每一項都需要大量的資料計算與決策判定。為了讓機械手臂的運作更有效率，本研究運用 affordance 的概念，希望利用這些資訊和機械手臂過去學習的經驗，簡化決策判定的過程，提升整體操作的效益。使用 Ontology 的概念，把物體、環境、任務的特性以及彼此之間的關係做連結，接著利用這些關聯性和特性，使用階層式任務網路規劃(Hierarchical Task Network planning)規劃出一連串合宜的動作。並於實驗階段使用本校實驗室的機械手臂，驗證加入 affordance 的概念後，對於環境、物體、工具等的認知有所差異，因而讓手臂能執行合適的方案，順利完成目標物抓取和搬移的任務。
Robot grasping belongs to the process of controlling robots, which includes environmental detection, subject identification and pathway planning and requires massive data calculation and strategic decision. The specific aim of this study is to use the information and robotic arm’s learning experience based on the concept of affordance to simplify the process of strategic decision and improve the efficiency of operation. By using the concept of ontology, we demonstrated the relation between the perception of objects, environment and actions. Then, we used Hierarchical Task Network planning to design a series of motion. We proposed the concept of environment-affordance, tool-affordance and object-affordance. At the end, we demonstrated some experiments to show their effects. Our system uses the information in a planner and executes the proper plan to complete the mission of grasping and moving through DOBOT Magician Robot Arm. The association between the tasks in a plan and the instructions to control the robot arm is explained
應用於光伏與壓電能量採集系統之單電感三輸入雙輸出升降壓直流-直流轉換器;A Single-Inductor Triple-Input Dual-Output DC-DC Converter for Photovoltaic and Piezoelectric Energy Harvesting Systems
No full text[[abstract]]本論文提出一應用於光伏與壓電能量採集系統之單電感三輸入雙輸出升降壓直流-直流轉換器。本電路以升降壓轉換器(Buck-Boost Converetr)做為單電感三輸入雙輸出(Single-Inductor Triple-Input Dual-Output, SI-TIDO)轉換器的基礎拓樸，並且設計為CCM/DCM 雙模式轉換器來提升運作範圍，然而共享電感會面臨輕重載模式難以判斷與不易補償等問題。本論文使用電流式PWM控制做為輕重載模式變換的依據，僅使用兩個補償器即在一個時脈週期內同步調整能量來源、儲能元件、與負載三者之間的功率，實現光伏與壓電雙來源能量採集系統，峰值效率達86.6%。其中最大功率追蹤的部分，本篇論文採用開迴路分壓(Fractional Open-Circuit Voltage)法來追蹤最大功率，使用電容分壓的方式實現比例常數，並搭配低漏電流取樣開關(Leakage reduction sample switch, LRSS)降低開關漏電，提升電壓維持時間。取樣電路於每1.048576秒取樣一次開迴路電壓，每次取樣花費256微秒。此外本論文提出一種延遲鎖定迴路(Delay Lock Loop, DLL)的零電流偵測器(Zero-Current Detector, ZCD)，其透過延遲鎖定迴路來鎖定比較器預開時間，因此相較於全時比較擁有較低的功耗，而相對於單點鎖定則有較快速的暫態響應。鎖定的時間定為100奈秒，因此擁有100奈秒的容許範圍。
This thesis presents a single-inductor triple-input dual-output(SI-TIDO) dc-dc converter for photovoltaic and piezoelectric energy harvesting systems. The SI-TIDO dc-dc converter uses buck-boost topology and can operate in continuous or discontinuous current modes that enhance the operation range of the converter. A new algorithm is proposed to share single inductor between all the inputs and outputs in one switch cycle. This algorithm determines light-load or heavy-load mode by current mode pulse width modulation control. Compared with conventional algorithms the proposed one saves a pair of compensator and mode detection circuits. Apart from regulating the output voltage to power the loading circuits, the converter also clamps the photovoltaic voltage to the maximum power point value. The fractional open-circuit voltage method is realized to track maximum power points by the capacitance divider circuit. Leakage reduction sample switch is used to extend the hold time of voltages. The peak efficiency of the proposed SI-TIDO buck-boost converter is 86.6%. The sampling cycle of the capacitance divider circuit is around 1.05 s. Each sampling takes 256 μs.Besides, a delay lock loop based zero-current detector(ZCD) is proposed. The proposed ZCD uses a delay lock loop to lock the time of pre-activation . Therefore, the power of proposed ZCD is lower full-time operation ZCD. The transient response of the ZCD is also faster than one point judgment ZCD because the ZCD have 100ns tolerance range
使用電壓掃描法應用於小型光伏面板之全域最大功率追蹤系統;A Design of Global Maximum Power Point Tracking Systems for Small Photovoltaic Panels Using Voltage Sweep
No full text[[abstract]]本篇論文完成一使用電壓掃描法應用於小型光伏面板之全域最大功率追蹤(Global Maximum Power Point Tracking, GMPPT)系統。雖然在周遭環境中，光能普遍存在且可以產生大量的能量，可是當太陽能板(Solar Panels, SPs)上發生部分遮蔽情形(Partially Shaded Conditions, PSC)時，會使最佳工作點產生飄移並使得功率-電壓電氣曲線呈現多個峰值的現象，導致提取的能量減少。因此，需求一個全域最大功率追蹤系統來追蹤多峰值中所對應的最佳工作點電壓，用以提高能量提取的效率。本論文所採用的全域最大功率點演算法是基於電壓掃描法。利用一功率級電晶體與運算放大器組成一個負回授迴路，只要給定參考電壓，便可以將Vpv穩在參考電壓值，接著透過步階方式來增加電壓便可以得到Vpv資訊，功率級電晶體上流過的電流為Ipv資訊，有此兩者資訊並利用電路實現演算法便可以得到最大功率點電壓。為了避免取樣電容漏電效應，在此將類比資訊轉為數位碼進行運算，並透過D型正反器儲存數位碼。最終透過數位類比轉換器(Digital-to-Analog Converter, DAC)將最大功率點數位碼轉成類比電壓。追蹤至全域最大功率點的時間約為10ms，系統待機時間為10s，全域最大功率點電壓追蹤峰值效率為99.38%。最後，本實驗建構一小型並列太陽能板，透過兩塊不同規格的太陽能板並聯而成，VOC分別為2.5V與4V，尺寸分別為60×65 mm2與60×110 mm2。本論文晶片使用台灣積體電路公司0.18μm 1P6M CMOS製程，以48 S/B封裝，尺寸為2.1×1.8 mm2。
This thesis completes a design of global maximum power point tracking systems for small photovoltaic panels using voltage sweep. Although light energy is ubiquitous and can generate large amounts of energy in the surrounding environment, however, when the partially shaded conditions occurs on solar panels, the best working point is drifted so that the P-V electrical curve presents multiple maxima, resulting in reduced energy extraction. Therefore, a global maximum power point tracking system is required to track the optimum operating point voltage corresponding to multiple maxima to improve the efficiency of energy extraction.In this thesis, the algorithm uses voltage sweep method. The use of power transistor and opamp to form a negative feedback loop, as long as a given reference voltage, Vpv can be stable to reference voltage, and then through the step to increase the voltage can get information of Vpv, current of power transistor is the information of Ipv, have both the information and use the circuit to achieve the algorithm, can get the maximum power point voltage. In order to avoid sampling capacitor leakage effect, the analog information will be converted to digital code to operate, and through the D-type flip-flop to store digital code. Finally, the digital-to-analog converter converts the digital code of maximum power point into analog voltage. The time to track to the global maximum power point is about 10ms, the system standby time is 10s, and the tracking peak efficiency of the global maximum power point voltage is 99.38%.Finally, parallelly connected solar panels with different specifications were constructed in the experiment. VOC are 2.5-V and 4-V, respectively, and the dimensions are 60×65 mm2 and 60×110 mm2, respectively. The chip is implemented by Taiwan Semiconductor Manufacturing Company (TSMC) 0.18μm 1P6M CMOS mixed-signal polycide process. The package of chip is 48 S/B and the die area of the chip is 2.1×1.8 mm2
用於含太陽能與蓄電池儲能之社區型微電網住宅型用戶電能消耗評估之模型與控制方式開發;Development of the models and controls of community microgrids with PV and battery energy storage for the assessment of residential-type users’ electric power consumption
No full text[[abstract]]國內電能消耗主要可分為工業電力、農業電力、商業電力及住宅電力等，影響能量消耗主要為經濟發展、氣候因素及人口規模等因素，使得電能消耗量逐漸上升。在經濟上，發展大型重工業皆需消耗大量之能源，電能消耗量也隨之增加。人民生活日常上也與電能消耗息息相關，以一般住宅用戶來說，因人口逐漸增加及天氣改變，導致家電用品之需求(如空調)提升，使得電能消耗節節上升。由於一般住宅之電能需求皆由傳統能源(市電)提供，但傳統能源與住宅屬單一電力潮流，如果市電端無法藉由升級系統來因應逐漸增加之負載需求時，可能導致住宅用戶面臨限電或停電危機。因此藉由引進社區型微電網系統，使其與市電端共同對用戶負載供電，可降低上述問題發生之可能。再則，社區型微電網由再生能源發電、儲能系統及控制系統組成，可監測負載、市電及再生能源與儲能系統之電量，且在其運轉下用戶能有效獲得較經濟的電能供應。在微電網技術研究中，為節省在實體系統進行測試所需耗費之成本，以及減少因技術不確定性所造成之不便，可透過系統模擬之方法來驗證，相關方法技術實現在真實系統之可行性。本論文以國內實際社區型微電網系統為範例，首先建立包括再生能源設施、儲能系統、市電端與用戶端網路及電能轉換器控制器等模型，並提出一協調控制策略。緊接，採用即時模擬技術來解決以往離線模擬所產生之相關限制。再則，透過全系統模擬，在併網、孤島與限電等運轉模式與情境下進行電能模擬分析；此外本論文也加入負載預測之設計，增加模擬系統之功能性。最終，經由所建立之模擬系統用於社區型微電網住宅用戶電量評估，相關模擬結果除可看出社區型微電網運轉所帶來之電費節省效益，更能驗證本論文所提建模方法與控制策略之有效性。
Domestic energy consumption is mainly divided into industrial power, agricultural power, commercial power and residential power, etc. Economic development, climatic factors and population size are common factors which influence the energy consumption and bring on gradual increase of energy consumption. In the economy, the development of heavy industry is required to consume so much energy that the energy consumption increases as well. People's daily life is also closely related to energy consumption. For the general residential users, the gradual increase in load demand will lead to power outage crisis. Therefore, the above problems can be reduced after the community-based micro-grid which is composed of renewable energy generation, storage and control system incorporated into the system. Furthermore, the community-based micro-grid not only can monitor the load demand and power supply but also it can save customers money and utilize energy more efficiently at the same time. To save the cost of testing on physical system, we could verify the feasibility of the proposed method through the system simulations.This thesis analyzes the cases based on actual community-based micro-grid system with construction of renewable energy sources, storage system and controller models and proposes some controlling strategies. Moreover, real-time simulation techniques are used to resolve limitations of off-line simulations and simulation analysis is implemented in condition of grid mode and island mode and limiting power, etc. In the thesis, load forecasting is also executed to extend the functions of simulation system. With the implementation of system simulations, the results show that it not only brings economic benefit for customers but also validate the efficiency of the proposed methods and controlling strategies
運用資料探勘技術預測臉書醫療健康粉絲團活躍率之研究;Research on Predicting the Active Rate of Facebook Fan Page of Medical Health by Using Data Mining Technology.
No full text[[abstract]]自從臉書推出粉絲團的功能之後，有越來越多的粉絲團成立，這也包含醫療院所成立的粉絲團。許多研究指出在Facebook粉絲團上鼓勵用戶互動、參與，提升用戶滿意度，可以有效增進用戶對品牌的忠誠度。醫病之間的溝通也使用粉絲團，只要能在粉絲團上多多促進、鼓勵用戶的參與，就可以提升用戶對品牌的忠誠度，進而對醫療院所的忠誠度也會跟著增加。 本研究使用Facebook Graph API抓取1,706個臉書粉絲團的貼文、留言等相關資料，再進行彙整、統計，並且以C4.5決策樹、Random forest、支援向量機、邏輯斯迴歸等資料探勘的技術找出關鍵因子。 本研究的建議是：提高貼文頻率，並以照片、影片貼文為主，還要注重貼文反應，要注意訪客貼文，不要讓自己的粉絲團變成廣告園地，不要讓負面情緒瀰漫。
Since the launch of Facebook fan page, many fan pages sited up , which includes Medical fan pages.Many studies have pointed out that Facebook fan page to interaction, participation, enhance customer satisfaction can effective enhance the user's Brand loyalty.In this study, the Facebook Graph API was used to capture the information of post and comment of 1,706 clinic fan pages, and then the data were collected. The data were analyzed by C4.5 decision tree, Random forest, support vector machine and logistic regression to identify key factors.The recommendations of this study are: To improve the frequency of photos or video posts, to care the reaction of posts, to care the customer posts, do not let your fan page become advertising garden, do not let the negative emotions filled
建立非計畫性14天內再入院之預測模式 ：以南部某區域教學醫院胃腸肝膽科為例;Establish a prediction model of the unplanned readmission patient after leaving hospital within 14 days ,take Gastrointestinal Hepatobiliary division for example.
No full text[[abstract]]病人短時間內反覆住院，不僅加重醫療成本的支出，也影響病患完善醫療的照顧與權益，更增加家屬的心理及經濟負擔。雖有少數非計畫性再住院是無法避免的，但仍應努力減少。本研究以南部某區域教學醫院胃腸肝膽科住院病患為對象，收集2011年至2013年期間再入院病人之各項社會人口學及臨床特徵和住院處置與出院後照顧資料，含性別、年齡、社會季節、飲食習性、高血壓、消化系統及其他疾病、住院和出院計畫、出院準備、後續衛教、居家護理、家屬支持照顧等，以決策樹及邏輯斯回歸之分類技術，建立胃腸肝膽科出院後14天再入院之預測模式，期有效的減少再住院率，提高醫療品質。
Repeated admissions of patients in a very short period of time not only increase the medical costs and expenses, but also has a great impact on the quality of medical service and family’s burden. Except for a very few exception, unscheduled admissions should be avoided whenever possible specifically of those re-admissions that are within 14 days after the last discharge. With information collected during the admissions at the department of gastroenterology of a regional hospital in south Taiwan from 2011 to 2013, this study aims to test predictive models hypothesized for their re-admissions against various socio-demographic, clinical and other supportive factors using multiple logistic regressions