透過您的圖書館登入
IP:18.206.185.68
  • 學位論文

八人制拔河團隊坐地後輪流起身之類神經網路模式研究

The Study of Neural Network Model on Ordering Raise After Sitting

指導教授 : 王金成
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


八人制拔河團隊坐地後輪流起身之類神經網路模式研究 研究生 翁梓林 指導教授 王金成 博士 摘 要 本研究目的在於建構一拔河團隊坐地後起身動作之類神經網路模式,以表達拔河團隊坐地起身之動作技術,以利往後進行各種不同輪流起身方式之模擬研究。受試對象為台灣地區大學女甲組選手共八名,身高165.8±5.5公分、體重61.1± 6.3公斤及年齡18.8 ± 0.7歲。根據實驗之設計,模式包含模擬單人起身分解動作、兩人組合輪流起身動作及三人任意輪流起身之動作等三種不同方式之人數組合,然後再作為團隊拔河坐地後輪流起身之主要模擬依據。起身之動態性動作中以高速攝影機(JVC-9800, 60Hz)沿受試者矢狀面進行2D拍攝,將拍攝之動作以APAS動作分析系統取得下肢段關節角度之運動學參數;並透過Bio-vision多功能測力系統(900 Hz)獲取拉力值。運動學與動力學資料處理乃透過自行輯寫程式以同步之方式擷取坐地後起身動作之拉力值,與對應之髖、膝及踝關節角度等變化,並以Lagrange之內插法(interpolation)給予百份等份之標準化,作為倒傳遞類神經網路模擬團隊拉力之訓練及測試範本依據。研究結果發現模擬單人分解動作之拉力、兩人輪流起身組合動作之拉力及三人坐地後任意輪流起身動作之拉力,在實際值與模擬值之誤差百分比率皆不超過百分之五。 根據模擬三種不同方式人數組合之起身動作具有良好精確度,進一步模擬團隊八人輪流起身動作之合拉力,並以估計標準誤檢驗類神經網路模式之精確度,結果發現在實際值與模擬值之誤差百分比率介於百分之四至百分之九。因此,以類神經網路之理論所建構模式,能有效表達團隊拔河坐地後輪流起身過程之拉力,此模式建立不僅可達單純化之目的,同時亦具有良好精確度。 關鍵詞:倒傳遞類神經網路、坐地後輪流起身動作、拔河

並列摘要


The Study of Neural Network Model on Ordering Raise After Sitting in the Tug-of-War ABSTRACT The purpose of this study was to establish a neural network model for simulating on ordering raise techniques after sitting in the tug-of-war. The subjects participated in this study were the university elite tug-of-war players with age 18.8 ± 0.7 years, height 165.8±5.5 cm and weight 61.1±6.3kg in eight girls. In this experimental design, the model contained three different combinations of subjects: (1) one individual subject, who performed the four separate motions, (2) two-subject combination, and (3) three-subject combination. This simulation served as the basis of ordering raise after sitting in the tug-of-war. JVC-9800 high-speed camera (60 Hz) was used to 2D cinematograph analysis at sagittal plane, and load cell of Biovision feedback system (900 Hz) to collect data when the subject raised movement after sitting during Tug-of-War. By processing the analysis of joint’s angle of lower extremity and resultant force, which was determined by back-propagation nerural network model to estimate ordering raise after sitting in the Tug-of-War. And based on the results, the standard error of estimate of values between experiment and neural network model was 25.58 kgf to 42.55 kgf which was no more than 9 % of the percentage of error. Therefore, the neural network model established by this study could provide a good and high accuracy simulation of the resultant force of ordering raise after sitting in the tug-of-war. Key words: Tug-of-War、Neural Network Model、Ordering Raise After Sitting

參考文獻


慶祝建國七十年體育學術研討會專刊。
許樹淵。(民86)。運動生物力學。國立編譯館主編。臺北:合記
朱文光。(民70)。拔河姿勢之研究。中華民國大專院校體育總會
彈性模式探討。中華民國大專院校體育總會八十六學年
華民國體育學會。中華民國體育學會九十年度學術論文發

被引用紀錄


劉家呈(2007)。八人制拔河團隊隊形比賽起始階段之運動學分析〔碩士論文,國立臺灣師範大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0021-2910200810534317
翁子傑(2012)。2010世界盃女子拔河冠軍隊選手起步動作之動力學分析〔碩士論文,國立臺灣師範大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0021-1610201315312273

延伸閱讀