基於導數尖波及型態學方法提取有效想像運動腦波;Extract effective imaginary motion brain waves based on derivative spike and morphology method

NCU Institutional Repository > 資訊電機學院 > 電機工程研究所 > 博碩士論文 > Item 987654321/86698

jsp.display-item.identifier=請使用永久網址來引用或連結此文件: http://ir.lib.ncu.edu.tw/handle/987654321/86698

题名:	基於導數尖波及型態學方法提取有效想像運動腦波;Extract effective imaginary motion brain waves based on derivative spike and morphology method
作者:	黃于洲;Huang, Yu-Chou
贡献者:	電機工程學系
关键词:	型態學法;動態時間規畫法;機器學習分類器;導數有效想像運動法;腦機介面;morphology method;Dynamic time warping;machine learning classifier;derivative effective imaginative movement method;brain-computer interface
日期:	2021-08-16
上传时间:	2021-12-07 13:07:48 (UTC+8)
出版者:	國立中央大學
摘要:	本論文基於導數有效想像運動法及型態學法來針對原始想像運動做有效區間判斷，並利用動態時間規畫法來進行有效想像運動時間點的驗證。會利用腦波試驗段中樣本數的特性做有效想像運動開始點的判斷，根據訊號的形狀，峰值，維持時間，通道間訊號的關聯性來進行有效想像運動演算法的計算。將原始腦波訊號做完資料擷取的部分後，便送入三種機器學習分類器中分別做想像運動辨識率的判斷，目的是嘗試觀察是否會因不同分類器而造成不同效果回饋。最後判斷的分類效果才能成為有效命令給予腦機介面做判斷。資料集的部分在公開資料集中提升辨識效率達到5.61%，而在非公開資料集的部分提升辨識效率達到5.93%，因此不僅可以有效降低資料樣本數的個數，亦可以使最後判定辨識率達到優化的效果。 ;The thesis is based on the derivative effective imaginary movement method and morphology method to judge the effective interval of the original imaginary movement. Moreover, the thesis uses the dynamic time warping method to verify the effective imaginary movement time point. The characteristics of the number of samples in the brain wave test segment will be used to determine the starting point of the effective imaginary movement. Then, the effective imaginary movement algorithm will be calculated according to the shape of the signal, the peak value, the maintenance time, and the correlation of the signal between channels. Once the original signal (Electroencephalography) is collected from the data, three machine learning classifiers can be used to judge the classification rate of the imaginary motion respectively. The purpose is to observe whether different classifiers will cause different effects. The classification effect of the final judgment can become a valid command for the BCI (Brain Computer Interface) to make judgments. The part of the data set improves the identification efficiency in the public data set about 5.61%; and the part of the non-public data set improves the identification efficiency about 5.93%. Therefore,it not only effectively reduces the number of data samples , but also optimizes the final identification rate.
显示于类别:	[電機工程研究所] 博碩士論文

文件中的档案:

档案	描述	大小	格式	浏览次数
index.html		0Kb	HTML	98	检视/开启

在NCUIR中所有的数据项都受到原著作权保护.

社群 sharing

数据加载中.....