以數位訊號處理單晶片實現可攜式即時睡意辨識系統

根據統計資料顯示，疲勞駕駛是導致交通事故主要原因之一，疲勞駕駛最容易造成精神狀態不佳以致產生睡意，然而在這種情況下高速行駛，很可能導致生命危險的嚴重事故。而腦波（electroencephalogram，EEG）訊號對於警覺度來說是相當重要的資訊，在以往的腦波警覺度分析大多是以電腦進行離線處理，缺乏即時處理功能。因此本研究基於腦波訊號建構一個可即時偵測駕駛者睡意之系統，透過本研究設計六通道主動乾式電極（active dry electrodes）以非侵入式擷取腦波訊號，並以TMS320VC5510 DSP單晶片為運算核心，再搭配MSP430F149主控整個系統的流程，以達到可攜式即時系統的目的。在腦波訊號處理方面，首先以穩態小波轉換（stationary wavelet transform，SWT）拆解，接著以積分值與零交越點兩種方式強化腦波訊號的特徵，再由倒傳遞類神經網路（back propagation neural network，BPN）進行警覺度分類。此系統可分辨清醒與睡意兩種精神狀態，當偵測出睡意時會發出警訊提醒駕駛者以減少意外的發生。經實驗測試，清醒辨識率可達到79.1%，睡意辨識率為90.91%。另外，本自製乾式電極系統優於以往的腦波量測電極，可穿過頭髮輕鬆的量測到腦波訊號，希望將來可以取代傳統的腦波電極應用於一般日常生活中。

關鍵字

乾式主動電極； MSP430F149 ；腦波； TMS320VC5510 DSP ；睡意；穩態小波轉換；倒傳遞類神經網路

並列摘要

The existing statistical data and survey reports indicate that the driver’s fatigue is one of the major causes of traffic accidents. Fatigue can easily lead to drowsiness. However, driving in this situation may cause dangerous accident. Electroencephalogram (EEG) signals give important information about the vigilance states of any subject. In the past, EEG vigilance states used computer for off-line analysis and lacked real-time processing capability. Therefore, this study constructs a real-time EEG-based system for detecting driver drowsy. The proposed system uses a six channels active dry electrode system to acquire EEG non-invasively. In addition, it uses TMS320VC5510 DSP chip as the algorithm processor, and MSP430F149 chip as a controller to achieve portable real-time system. This study extracts two features of EEG signal: integral of EEG and zero crossings, and uses back propagation neural network to classify vigilance states. This system can discriminate alertness and drowsiness. When the system detects drowsiness, it will warn drivers against accident. The accuracy of the ANN is 79.1% for alertness and 90.91% for drowsiness state. Additionally, this custom-made dry electrode is better then traditional EEG electrodes. It can easily measure EEG through hair. We hope this dry electrode can replace traditional EEG electrode and applies in the daily life.

並列關鍵字

active dry electrodes ； stationary wavelet transform (SWT) ； MSP430F149 ； electroencephalogram (EEG) ； drowsiness ； TMS320VC5510 DSP ； back propagation neural network (BPN)

參考文獻

[24] 杜翌群, “以穩態小波結合PCA及ICA辨識手部動作肌電圖之評估”, 中原大學醫學醫學工程系碩士論文, 民國92年11月

[29] 陳建宇, “多電極式手部動作辨識系統”, 中原大學醫學醫學工程系碩士論文,民國90年6月

[30] 葛士豪, “即時手部動作辨識系統之實現”, 中原大學醫學醫學工程系碩士論文, 民國95年7月

[3] Conradt, R., Brandenburg, U., Penzel, T., Hasan, J., Varri, A., & Peter, J. H., “Vigilance transitions in reaction time test: A method of describing the state of alertness more objectively”, Clinical Neurophysiology, 110, 1499–1509, 1999

[5] Akerstedt, T., Kecklund, G., Knuttsson, A., “Manifest sleepiness and the EEG spectral content during night work”, Sleep 14, 221–225, 1991

國際替代計量

以數位訊號處理單晶片實現可攜式即時睡意辨識系統

全文下載

主題瀏覽