Chung Cheng University Institutional Repository
Not a member yet
889 research outputs found
Sort by
資料導向非平衡型分類於推薦系統之研究;A Data-Driven Imbalanced Classifying Method for Recommendation Systems
No full text[[abstract]]推薦系統之主要目的是提供使用者可能喜好的項目或服務。但是要進行推薦時，因受限使用者歷史評價紀錄中，每位使用者曾經評價過的項目，實際僅占可選擇之全部項目的極少比例，導致使用者評價矩陣呈現稀疏狀態。當系統能用於分析的評價資料過少，就會使推薦的難度提升。也因為稀疏問題會有推薦成效不佳的影響，所以過去學者們也一直提出新的方法及架構，嘗試提升推薦準確度。本論文是以參考使用者的歷史評價紀錄做為推薦的依據，故採取評價計算的協同式過濾方法，而MovieLens 電影資料庫會做為我們的實驗資料。在改善推薦系統典型的稀疏問題，則是採用重採樣技術(SMOTE)做為應對方針，它主要是用於看待非不平衡型數據資料的一種處理方式，我們將它放在推薦系統處理評價資料的過程中，藉由增加虛擬評價樣本到訓練集合，並建立成預測模型來找出使用者感興趣之電影。最後，本論文所提供的方法可以提升電影推薦系統的預測準確率。
The main goal of a recommendation system is to suggest services or items that users like. However, the system suffers from the sparsity problem if the percentage of user feedbacks is very small in the historical database. The sparsity problem will have the effect of poor recommendation since a small number of the contributed user ratings are too difficult to analyze. Therefore, in order to improve the accuracy of recommendations, many researchers have referred to different methods and frameworks. In this thesis, we try to incorporate an oversampling method to increase the number of useful data. The MovieLens database is used as the experimental data in our study, and we add the resampling technique, the Synthetic Minority Over-sampling Technique (SMOTE), which is a way for processing imbalanced classification problem in our recommendation system. While training dataset is preprocessed by the sampling method, the new training dataset will be built as a prediction model to find movies that the user is interested in. Finally, we show in our experiments that the proposed method in this thesis can improve the prediction accuracy of the movie recommendation system
傳輸線模擬電路-可肉眼觀察射頻訊號在傳輸線中的行為 ;Transmission Line Simulation Circuit-obseving RF signals’ action in transmission line
No full text[[abstract]]本論文研究主題-傳輸線模擬電路,此電路為能表現出射頻訊號在傳輸線中的行為,可看見暫態的行進與干涉共振後的現象,因射頻訊號特性複雜,在相關領域的知識多用數學式來表達與推算,因此概念抽像不易理解。針對抽象問題,且不同於市面的軟體模擬,本論文討論為用硬體技術來完成用肉眼就能觀察的功能,且最後也有課程的實際應用和問卷回饋,得不錯的回應,也證明此傳輸線模擬電路研究確實有實用價值。
This paper contains one subject-transmission line simulation circuit. This circuitcan show RF signals’ actions in the transmission line and make observers see thesignal moving or resonating in naked eyes. Because RF signals’ characteristics arecomplex and usually described with arithmetic formulas, related concepts are abstractfor understanding. To improve understanding difficulty, this paper use hardware todesign the fuction of observing in naked eyes. Moreover, there is a course and aquestionnaire for the application of this research. Then getting a great result provesthis transmission line simulation circuit having an applicabilit
寬頻具濾波效果之功率分配器與寬頻帶通濾波器之設計;Design of A Microstrip Filtering Power Divider With A Wide Passband and Broad Stopband and Wideband Bandpass Filter
No full text[[abstract]] 本論文分成兩個部分，第一部分介紹寬頻及寬截止頻之功率分配器，在此所提及之功率分配器首先是由兩段四分之一波長之耦合線以及一開路殘段形成一多模態諧振腔做為出發，以此架構可得到寬頻的效果，之後為了再加以匹配，在輸入埠加入L-type傳輸線加以匹配至最佳狀態，最後再加入電阻，改善兩輸出埠之間的隔離度，最後為了要達到寬諧波抑制，將前段之四分之一波長的耦合線，用三段等長之步階式耦合線來取代，以達到諧波抑制之功效。 而在第二部分介紹寬頻帶通濾波器，利用四個類環型諧振腔來達到寬頻的特性，而藉由適當的選取每個諧振腔之電氣長度，可將諧波往後推移，利用耦合係數法來進行設計，可計算出類環型諧振腔之耦合間距。
The thesis includes two parts, the first part will introduce the microstrip filtering power divider with a wide passband and broad stopband. In this part, I use quarter-wavelength coupled line and a open-stub to be a muti-mode resonator. This structure can be a broad band performance. In input port, I use L-type transmission line(TL) to match this circuit and add resistor to improve isolation of two output port. Finally, we want to achieve harmonic suppression. I use three equal length stepped couple line to replace quarter-wavelength coupled line at input port. In the second part, I will introduce the wideband bandpass filter at chapter 4. By using four pseudo-hairpin-line resonator to achieve broadband performance and using coupling coefficient method to get coupling gap of each resonators
以倒傳遞及卷積神經網路辨識心律不整心電圖之研究;Back-propagation and Convolutional Neural Networks for Arrhythmia Electrocardiogram Classification
No full text[[abstract]]心電圖辨識(Electrocarfiogram discrimination)在臨床診斷不同心臟疾病方面扮演著相當重要的角色。目前已有很多關於心電圖辨識的方法被提出來，在許多議題上仍有很多空間可以加以探討與改善。本論文我們考慮以留一病患驗證來探討倒傳遞類神經網路(Back-propagation neural network, BPNN)與卷積神經網路(Convolutional neural network, CNN)的辨識率，並建構一套辨識心律不整的系統。本篇論文方法上使用離散小波轉換將訊號分解成不同次頻帶(Subband)成分，接著使用高階統計(Higher order statistics)來描述心電圖訊號，用以提升心搏辨識過程中抗雜訊的能力，另外再加上RR區間的相關特徵，期望對辨識能有所幫助，分類器方面則採用BPNN與CNN來探討。提出的辨識方法分為三個部分，第一個部分為利用BPNN並以排除個體差異的方式驗證，第二個部分為利用BPNN並以留一病患來驗證，第三個部分為討論BPNN與CNN的相關性來結合其優勢並分類。從研究結果得知，利用BPNN做排除個體驗證並搭配2-Folds可達到95.52%；以留一病患驗證方面，正確率僅有51.4%，其原由為每個心律不整的心搏數相對少於正常心搏，造成資料不對等而分類偏向於一類的情形，有鑑於此，本研究提出雙層分類、加入個體參考訊號、加入CNN擷取的特徵和改良BPNN的初始權重值等方法，嘗試提升留一病患驗證時的正確率，結果顯示，當採用雙層辨識架構時，其辨識率最佳可達91.89%，相較於直接分類辨識，其正確率提升超過40%。
Electrocardiogram discrimination plays a very important role in the clinical diagnosis of heart diseases. A lot of methods for ECG beat classification have been proposed. However, many topics still leave room to be improved. In this paper, we build arrhythmia recognition systems based on back-propagation neural network (BPNN) and convolutional neural network (CNN). We will focus on the accuracy of leave one patient out.In this paper, discrete wavelet transform was used to decompose the signal into different subband components, and then higher order statistics were used to describe the ECG signal to enhance the noise-risisting ability. RR-Interval related features were further included to improve the accuracy. Finally, the beat types would be classified by BPNN and CNN.The proposed method is divided into three parts. First, the beat types are classified by BPNN and the validation method is excluding individual differences. Second, the beat types are classified by BPNN and the validation method is leave one patient out. Third, we discuss the relevance of BPNN and CNN and try to combine their advantages to classify the beat types. The results shows that the accuracy by BPNN with 2-Folds validation method is 95.52%. However, while using the leave one patient out validation, the accuracy is 51.4%. Since the arrhythmia beats are far less in number than the normal beats, this phenomenum causes bias in classification and leads to low accuracy. In order to solve these problems, we proposed a two-level classification scheme by adding reference signals of the test sample, adding the features extracted from CNN, and modifying the initial weight of BPNN intending to improve the accuracy of leave one patient out validation. By using the two-level scheme, the best accuracy can reach 91.89% with leave one patient out validation. Compared with the direct classification, the improvement of accuracy is more than 40%
考慮氣象組合模型之風力發電機率預測研究;Study of Probabilistic Wind Power Forecasting Considering Weather Ensemble Models
No full text[[abstract]]近十幾年來，隨著再生能源占比日益增加，為了降低電網的經濟成本並使其擁有一定的可靠性，「風力機率預測」成為問題決策的必要考量因素。機率預測與傳統決定性預測的最大差別在於機率預測會額外把預測的不確定性一起納入考量，而這對再生能源發電預測而言是相當重要的。本論文將先對幾種常見的機率預測做整理，並詳述這些方法的概念與預測方式。接著，本文整理評估機率預測的方法。最後，本研究提出非參數風力發電機率預測模型。此模型採用上下限估計法，從輸入資料中預測出可靠的風力預測區間；這些輸入資料包含風場的歷史風力與風速量測資料，以及台灣交通部中央氣象局所提供的20組數值天氣風速預報值。其中，上下限估計法中的參數是由電荷搜尋演算法調整而得。本文將用三個台灣實際風場的發電資料進行預測模型的比較。研究結果指出本文提出的模型皆能比持續法的模型有更佳的表現。
Over the past one to two decades, probabilistic forecast of wind power generation is considered as a necessary input to decision-making problems for reliable and economic power systems operations, especially when the penetration of renewable energy started to grow rapidly. Probabilistic forecast differs from traditional deterministic forecast in that it also takes uncertainty into account.This thesis provides an overview of state-of-the-art technologies on wind-power probabilistic forecasts, and describes the fundamental concepts of these forecasting methods. Then, various methods are summarized to evaluate the performance of a probabilistic forecast model. Finally, this work also proposes a modified nonparametric method for construction of reliable Prediction intervals (PIs). Lower upper bound estimation (LUBE) method is adapted for construction of PIs for wind power generation considering the ensemble wind speed data provided by the numerical weather prediction (NWP) system of the Central Weather Bureau (CWB) of Taiwan. Charged search system (CSS) is employed to adjust parameters in LUBE. Performance of the proposed method is examined using data sets taken from three wind farms in Taiwan. Simulation results indicate that the quality of PIs from proposed model is significantly superior to the quality of PIs constructed from persistence model for one-hour-ahead time horizon
雙饋感應發電機在機械暫態的改良式通用模型之建立;Development of a Modified Generic Model of a DFIG-based Wind Turbine Considering Mechanical Transient
No full text[[abstract]]近十年來，風力發電在電網的占比大幅地增加。如此大量的風能併入電網，讓電網的操作人員在維持電網的運轉穩定性這一方面，面臨了諸多挑戰。為了模擬及分析大量風能併入大型電力系統所造成的影響，風力發電機模型的結構及準確度就變得非常重要。若將風力發電機的複雜模型應用於電力系統的穩態運轉及動態穩定度分析，需要花費較長的計算時間，也會加重電腦計算的負擔。因此，擁有簡化的結構及適度的準確度之通用模型已經被提出，用以模擬各種類型的風力發電機。本研究的風機通用模型及複雜模型是根據目前市面上最普遍的雙饋式感應風力發電機所建立的。為了確保本研究建立的風機模型能夠模擬實際風機的電氣及機械特性，傳動軸以及含抗飽和功能的槳距角控制必須被納入模型。此外，本研究亦考慮風速變動造成的機械暫態，以確保風機模型的完整性。本研究將風機的複雜模型視為參考標準，進行通用模型及複雜模型兩者之間的比較。本研究的模擬結果證明在調整通用模型的數個重要參數之後，通用模型和複雜模型兩者的動態響應及特性會非常相近。
Over the last decade, the wind power penetration has increased substaintially because of the environmental consideration. With the integration of so much wind power generating capacity, the system operators are facing the challenges of maintaining the power system stability. Thus, the structure and accuracy of the wind turbine (WT) model are critical to the simulation and analysis of the large-scale power system. The use of the complex model to analyze the steady-state operation and dynamic stability of the power system requires long computation time and heavy computation burden. Accordingly, the generic models of the various WTs with the simplified structure and the adequate accuracy have been developed. This work develops both generic and complex WT models that are based on the doubly-fed induction generator (DFIG). The DFIG-based WT, which is the so-called type-3 WT, is the most popular in the world nowadays. To ensure that the developed generic and compex models can represent both electrical and mechanical characteristics of a real WT, the drive train and the pitch control with the anti-windup mechanism are included. Also, the mechanical transient due to the wind speed change is considered to make the both models more complete. This work compares the developed generic model with the developed complex model. Considering the complex model as the reference, several parameters of the generic model are adjusted. The simulation results demonstrate that the developed generic model of the type-3 WT reasonably approximates the developed complex model
太陽能換流器之控制策略與諧波抑制研究;Research on Control Strategy and Harmonic Suppression of PV inverter
No full text[[abstract]]本論文主要以三相太陽能併網發電系統的控制策略為研究對象，分析其操作原理並建立數學模型，針對太陽能電池模組的最大功率點跟蹤方法、併網發電系統的控制策略以及併網換流器得濾波器設計原理等問題進行各方面的探討。一開始本論文根據太陽能電池的電氣特性以及太陽能電池模組在不同的光照強度和環境溫度下的輸出特性，使用實際的太陽能模組的參數資料作驗證。也分別對目前幾種經典的最大功率跟蹤方法的動作原理作介紹。接著分析了三相太陽能併網換流器的工作原理和拓墣架構，以並網換流器的濾波器為切入點，建立三相太陽能併網換流器的狀態空間數學模型，以利於第四章研究不同的三相太陽能併網換流器控制方法。在控制方法上，分別針對PI和PR控制器的原理，透過應用至換流器中的控制架構與轉移函數對其作分析，並根據PR控制器的特性介紹的新型的准PR控制器。准PR控制器也能應用在抑制電網諧波，然而傳統架構上的准PR控制器諧波補償效果並不佳，因此提出了一種改良式的准PR控制器諧波補償架構。 最後在Matlab/Simulink中建立了三相太陽能併網發電系統，並透過模擬驗證前述PI和PR控制器的特性，以及其併網效果。也驗證了改良式的諧波補償，比起傳統的補償架構有更良好的效果。
In this thesis, the control strategy of three-phase grid-connected PV power generation system is the main research object, and establish the mathematical model to analyze its principle. The maximum power point tracking method of solar cell module, control strategy of grid-connected power generation system and the designing principle of grid-connected inverter filter are also discussed.At the beginning of this thesis, according to the electrical characteristics of solar cells, using the actual solar module parameters to verify the output characteristics of PV array in different irradiation and ambient temperature. And also introduces the operating principle of several classic maximum power tracking method.Then, the working principle and topology of the three-phase grid-connected PV inverter is analyzed to establish the state space mathematical model. In the control method, through the application of the inverter control structure and transfer function to analyze the principle of PI and PR controller. According to the characteristics of the PR controller, a new type of quasi-PR controller is introduced. The quasi-PR controller can also be used to suppress the harmonics of the power grid. However, the harmonic compensation effect of the PR controller on the normal structure is not well. Therefore, an improved quasi-PR controller harmonic compensation structure is proposed.Finally, a three-phase grid-connected PV generation system is established in Matlab/ Simulink, and the characteristics of the PI and PR controllers are verified by the simulation. Also verified the improved harmonic compensation with a better effect on the harmonics compression which compared to the normal one
跑步間歇訓練組間不同休息時間對訓練效果影響之研究;Impact of training effect to different rest duration during high intensity interval training
No full text[[abstract]]目的: 探討長跑選手進行間歇訓練時，不同休息時間對於跑步速度、心跳率及運動後血乳酸的影響，並比較兩種間歇訓練的自選休息時間差異，以提供間歇訓練休息時間之建議。方法: 以國內8名優秀長跑選手 (6名男性、2名女性，年齡24.8±5.8歲、身高168.8±7.9公分、體重57.5±6.5公斤、5000公尺成績16.9±2.0 分鐘) 為受試對象，利用跑步機與Cosmed Quark CPET氣體分析系統進行前測，以取得VO2max、vVO2max及HRmax等相關數據。依平衡次序原則，將受試者分配成3組，每組皆進行4週，每週一次之1200M×5、跑步強度為vVO2max之測驗，間歇休息時間隨機分配為1、2及4分鐘及自選休息時間，共四次跑步漸歇訓練；另外再進行一次800Mx8、跑步強度略高於vVO2max、自選休息時間的間歇訓練，以比較不同距離間歇訓練自選休息時間之差異。結果: 間歇訓練時以1分鐘、2分鐘、4分鐘、自選休息時間 (1.94±0.68分鐘)，各趟次1200m的跑步運動時間、運動期心跳率及運動後血乳酸濃度皆沒有顯著差異，但休息一分鐘之休息期心跳率顯著高於其他休息時間。800mx8自選休息時間 (1.69±0.26分鐘) 間歇運動後血乳酸濃度顯著高於1200M×5自選休息時間。結論: 依據跑步時間、運動期心跳率、血乳酸濃度來看，進行1200 m×5跑步間歇訓練時，1-4分鐘休息時間具備相同的影響。1分鐘的休息時間會造成休息期心跳率偏高。
Purpose: The aim of this study is to investigate impact of training effect to different rest duration during interval training. And compare the difference of two self- selected rest durations to make suggestion for the runners who intents to do interval training. Method: 8 elite runners (6 male and 2 female, age 24.8±5.8, height 168.8±7.9cm, weight57.5±6.5kg, 5000M 16.9±2.0min) performed treadmill test to collect the VO2max, vVO2max and HRmax data as the reference of interval training intensity. All participants were assigned to 3 groups. Each group performed 4 sessions of 1200M x 5 rounds by vVO2max and 1 session of 800M x 8 slightly higher intensity than vVO2max of interval training on the track. The rest duration of each round was randomly assigned to each group by 1, 2 and 4 minutes in first 3 rounds of 1200Mx5. All participants performed the 4th session interval training by self-selected rest duration. The last session was 800Mx8 self-selected rest duration. Data was analyzed by One-way ANOVA, Repeated measure t test and Two-way ANOVA repeated measure to compare the maximum heart rate during running period, lowest heart rate during rest period, blood lactate at 3-5 minutes post exercise of each session and running time of each round of interval training in between different rest durations. The level of significance was set as .05. Result: The difference of 1200Mx5 running time, maximum heart rate during running period and blood lactate at 3-5 minutes post exercise of different rest duration 1 min, 2 min, 4 min and self-selected 1.94±0.68 min) were not significant. The lowest heart rate in 1 minute rest duration was significantly higher than other rest durations. The post exercise blood lactate of 800mx8 self-selected rest duration (1.69±0.26 min) was higher than 1200M×5 self-selected rest duration. Conclusion: According to running time, maximum heart rate during running period and blood lactate, the impact of 1-4 minutes rest duration would be same for1200 m×5 interval training. One minute rest duration for 1200Mx5 interval training would result in higher heart rate at rest duration
國中體育專家教師有效教學行為及其影響因素;Effective Teaching Behaviors and Influential Factors among the Experts Teachers of Physical Education in the Junior High School
No full text[[abstract]]本研究目的在瞭解國中體育專家教師有效教學行為及其影響因素。本研究採質性研究方式，以立意取樣方式，選取6位國中體育專家教師為半結構式訪談對象，其服務皆超過十年，教學熱忱、教學成效及口碑皆獲得肯定。另以文件分析及參與觀察等方式蒐集資料。將所得資料進行主軸編碼，持續歸納、比較及分析。結果發現國中體育專家教師有卓越的教學計畫、教學策略、教學評量、教學氣氛等行為，其有效教學計畫概念包括教學計畫、教學設計、內容與編排等；有效教學策略概念包括引起學習動機、多元教學、分組學習、示範講解等；有效教學評量概念包括評量內容、回饋與獎懲、教學反省等；有效教學氣氛概念包括教師熱忱、課室管理、師生關係、教師期待等。國中體育專家教師表現出卓越的教學計畫與準備行為，多元有效的教學策略與方法，精進有效的教學回饋、修正與檢討行為，以及良好的教學互動與師生關係。
The purpose of this study is to understand the effective teaching behaviors and influential factors among the expert teachers of physical education in the junior high school. Qualitative research method was adopted. Six expert teachers of physical education in the junior high school to conduct a semi-structured interview, who has serviced for more than ten years and has already gained a good reputation in teaching enthusiasm and teaching effectiveness. In this study, we will also collect data by document analysis and participation observation, and the information will be encoded, continuous induction, comparison and analysis. The results shows that experts teachers of physical education in the junior high school have excellent teaching plan, teaching strategy, teaching evaluation and teaching atmosphere. And their concept of effective teaching plan includes teaching plan, teaching design, content and arrangement. The concept of effective teaching strategies includes learning motivation, multidimensional teaching, group learning, demonstration; The concept of effective teaching assessment include assessment of content, feedback and rewards and punishments, teaching reflection; The concept of effective teaching atmosphere includes teacher enthusiasm, classroom management, teacher-student relationship, teacher expectation, etc. The results of the interview shows that the experts teachers of physical education in the junior high school have demonstrated excellent teaching plan and preparation behavior, multiple effective teaching strategies and methods, intensive and effective teaching feedback, revision and review behavior, and good teaching interaction with teachers and students
探討兩岸民眾對線上健康諮詢平台忠誠度之影響因素;The Impact Factors of The Taiwan and China People’s Loyalty on Online Health Consultation Platform
No full text[[abstract]]隨著社會經濟不斷的發展、生活水準的提高，民眾的醫療保健意識也正在發生變化。在目前的技術下，民眾可透過網路在足不出戶的情況下進行資訊、知識的蒐集與瞭解。而台灣衛生福利部及中國國務院為因應網路的趨勢及加強為民服務，分別推動線上健康諮詢平台，提供民眾上網獲得免費、專業的醫療諮詢，使民眾能透過網路平台，瞭解更多的醫療保健知識。但在兩岸都有線上健康諮詢平台的情況下，民眾對其的使用程度卻大不相同，在Toufaily等人(2013)文獻回顧消費者與網站忠誠度的研究中，歸納出信任、滿意度是決定忠誠度的重要因素，並提到先前對忠誠度這方面的研究較傾向於如何增加民眾使用網站的忠誠度，之後的研究可進一步探討文化差異對網路忠誠度的影響。因此，兩岸民眾在使用線上健康諮詢平台上使用率的差異，是否因兩岸文化不同的、民眾信任平台的程度等因素的影響，本研究將藉由常被運用來預測消費者對許多產品及服務的滿意度與忠誠度之期望確認理論加上信任變數，以文化做為調節變項，來探討兩岸民眾對線上健康諮詢平台的忠誠度之影響因素。在兩岸文化作為調節變項的狀況下，研究結果指出，民眾的自覺健康風險及使用平台後的確認程度越高，越會認為線上健康諮詢平台有用，而民眾對平台的確認、信任程度會影響其對線上健康諮詢平台的滿意度，當民眾認為平台有用及滿意時，其對線上健康諮詢平台的忠誠度越高。另外，本研究結果得知線上健康諮詢平台的有用性及滿意度間之關係，會受到文化內的不確定性規避構面所調節。綜合本研究結果，提升線上健康諮詢平台的有用性、增加民眾對線上健康諮詢平台的信任程度及忠誠度，並降低民眾對於未知的情況所感受到的恐懼程度，讓民眾能持續使用平台進而降低醫療資源浪費等問題。
With the rising of economy and living standards, people can now search for the information they want to know through the Internet at home. In response to this situation, Taiwan’s Ministry of Health and Welfare and the State Council in China promoted the online health consultation platform to provide people the free and professional medical advice, so that people can understand the knowledge of health through the platform. Both nations now have online health consultation platform, but the usage of each one is very different. Toufaily et al. (2013) summed up that trust and satisfaction are important factors to determine loyalty, and the previous studies on loyalty in this regard were inclined to show how to increase people’s loyalty about using such a site, suggesting that further research can explore the impact of different cultures on the web loyalty. Therefore, the difference between citizens in both sides, Taiwan and China, should be studied more to take into account their use of online health consultation platform, the influence of different cultures, trust and so on. This study is based on Expectation Confirmation Theory (ECT) and added trust to be the variables, and then used culture as the adjust variable to explore the influencing factors of people’s loyalty about online health consultation platform in both sides