| Summary: | 碩士 === 國立中興大學 === 電機工程學系所 === 102 === In recent years, accident news due to driver fatigue is often heard. When the driver appears nodded, sluggish, and inattention, it is better to enable an early warning to prevent the tragedy. This phenomenon often occurs in a long-time and high attention required condition. For example, long-distance bus drivers must maintain good physical condition in order to protect the lives and safety of passengers.
In general, the drowsiness detection warning systems use nod or blink frequency to make judgment. However, the above methods use the information that the driver has already entered a drowsy state. Hence, it can’t early warn the drivers. In this thesis, we use non-invasive measurement of electrocardiography (ECG) signal and heart rate variability analysis (Heart Rate Variability Analysis, HRVA) in the frequency domain. The results using statistic calculation are utilized to detect whether the driver enters the fatigue/drowsiness state or not. We can warn the drivers earlier by this method and it’s a good help to prevent traffic accident.
The thesis proposes an intelligent drowsiness detection warning system. The main study is ECG signal pre-processing, and drowsiness detection and warning. Enhanced So and Chan Method and Fast Fourier Transform are integrated together. The entire system is finally implemented on Zedboard. Graphical user interface (GUI) is developed under Linux Operating System (OS). It can immediately display ECG waveform and power spectral density (PSD) analysis of heart rate variability, and record each of the sympathetic / parasympathetic nervous balance index (LF / HF).
|
|---|
