結合慣性感測器姿態偵測與運動區腦波之腦波人機介面

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Electrical Engineering > Electronic Thesis & Dissertation > Item 987654321/77770

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/77770

Title:	結合慣性感測器姿態偵測與運動區腦波之腦波人機介面
Authors:	王昱凱;Wang, Yu-Kai
Contributors:	電機工程學系
Keywords:	慣性感測器;運動時腦波律動;腦波人機介面;Inertial movement unit;sensorimotor Mu rhythm;brain computer interface
Date:	2018-08-21
Issue Date:	2018-08-31 14:55:32 (UTC+8)
Publisher:	國立中央大學
Abstract:	自從大腦腦波訊號擷取的技術進步，科學家透過腦波訊號的擷取分析，協助神經或肌肉損傷的病患，運用腦波與環境溝通，讓腦機介面成為十分受到矚目的研究。而如何使系統能夠正確地判斷受測者所面臨或是正在執行的狀態，進而解讀腦波所隱藏的訊息。然而，腦波研究最關鍵的議題是如何獲得足夠的資料來對腦波訊號進行分類訓練，而正確的標記則是非常重要的關鍵議題。因此本研究希望藉由整合腦波訊號處理與慣性感測器系統，來提供足夠的腦波訓練資料，發展更貼近日常動作的腦波分析系統。本研究以慣性感測器做為工具，用於提供不同運動時腦波辨別的時間標記，目的在於結合慣性感測器與腦波機來提升腦波人機介面的準確率。我們在受試者雙手各裝置2個慣性感測，結合黏貼於C1、C2、C3、C4及前額位置的腦波，每隔5秒做一次手臂伸展動作，記錄受測者的動作姿態與腦波，透過肢體間的角度變化，來對腦波進行標記，標記方式為抓取準備要運動時的那一瞬間作為時間基準點，抓取的腦波資訊以此基準點向前取兩秒到向後四秒間的資料作為分析腦波的區間，我們只擷取動作時的腦波訊號做分析。透過與慣性感測器的結合測量腦波資訊，經過卷積神經網路(CNN)來取代以往傳統量測運動時腦波的方式。藉由慣性感測器的使用及CNN網路架構，找出動作與腦波相對應的連結。;Owing to the rapid developments in brain wave acquisition technologies, brain computer interface (BCI) has drawn great attention in recent years. Scientists are trying to develop BCI technologies for neuromuscular paralyzed patients to communicate with external environments through their intentions. Nevertheless, the key issue of achieving correct intention detection is the requisite of abundant brain wave data for classifier training. Especially, precise labeling of brain wave data in different conditions is important. Therefore, this study intended to combine EEG data with subject’s limb postures, in order to obtain training data for BCI from subject’s daily life movement recordings. In this research, we mounted two inertial movement units (IMU) on subject’s left and right wrists. The EEG electrodes were attached on C1, C2, C3, and C4 positions, according to international 10-20 EEG system. Subjects were asked to extend and flex their left and right elbows individually, and timing of subject’s posture data was wirelessly transmitted for EEG labeling. EEG data were segmented into epochs from -4s to 2s anchored to subject’s movement onsets. Labeled EEG data were used to train convolutional neural network for BCI detections. Our system has achieved acceptable detection rates by exploring the connections between subject’s movements and EEG changes.
Appears in Collections:	[Graduate Institute of Electrical Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	222	View/Open

社群 sharing

Loading...