An ECG-based Approach To Driving Fatigue Detection

Driver fatigue is a significant risk factor in transportation.It is useful to avoid traffic accident and keep driver's working efficiency for analyzing the physiological signals of drowsy and alert drivers. Of physiological signals applying in drowsiness-detection, the ECG signal was selected as determination of a driver's drowsy level.The thesis first introduces the history, the significance and the current research status of driving fatigue detection. In the end of this chapter, the main research content and method path of this thesis is proposed.Chapter two analysis the delay time and embedding dimension of periodic signal, random signal and ECG signal in the phase space reconstruction,and then study largest Lyapunov exponent, entropy and complexity of R-Peaks series and RR-intervals series. Through above analysis, a chaotic characteristic of R-Peaks series and RR-intervals series was anlysised, which is the basis for subsequent research.Chapter three does a research monitoring variation in linear and nonlinear parameters of ECG signal before and after an indoor simulated driving task. The experimental results showed that measurement of selected ECG parameters had gained great insight into the controlling mechanisms of homeostasis which provides an opportunity in the future work to quantify driving fatigue based upon degree of deviation from normal equilibrium states.Chapter four proposes kernel primary component analysis (KPCA) method to extract the features from normal and fault samples, in the course of fatigue classification, the method of PERCLOS was selected. The classification results showed that selected ECG parameters can be used as fatigue detection.Chapter five introduces the theory and arithmetic of support vector machine (SVM). After reviewing the one-classfication method of support vector data description (SVDD), the method was used to distinguish fatigue sample from normal sample. At the last, a ECG-based fatigue detection system was implemented, which verify the practicability and correctness of the theories, methods and technologies proposed in this dissertation.The last chapter summarizes the main work of the whole thesis and prospects the work which can be kept up.