Research On The Recognizing Arithmetic Of Fatigue Driving Based On The Physiologic Factor Analysis

Fatigue driving is one of the main reasons for tragic traffic accidents and we identify the importance of developing driver fatigue countermeasure devices in order to help prevent driving accidents. Numerous physiological indicators are available to describe an individual's level of alterness such as electroence-phalogram(EEG), heart rate, etc. We assessed four subjects during a driver simulator task with the aim to isolate 5 specialed physiological indicators, including heart rate(HR), skin conduction(SC), electromyogram(EMG), temperature and respiration, changes during normal, medium and extreme phases of fatigue during driving. From these collected data, some physiological eigenvalue characterized the subjects' state of mind was abstracted and a physiological information database was developed. Using these date we subsequently suggest that the subjects' drowsiness can be recognized by fuzzy C-Means arithmetic.The fatigue driving experiments were carried out on the car simulator. The subjects' physiological data was acquired by ProComp Infiniti, a multi-modality system for real-time computerized biofeedback and data acquisition. The acquisition data included HR, SC, EMG, skin temperature, respiration rate and respiration amplitude. Before the experiments, an orthogonal experimental design has been made and the sex, the time the experiment made and the driving environment are consider as factors that may affect the state of the mind of the subject. According to the state of the mind of the subjects during the whole experiment, the conclusion of which factor affects the state of the subject most can be drawn by the method of range analysis.On the basis of the acquisition data, 16 groups data of each state, including normal, medium and extreme phases, are abstracted and analyzed. While analysis results in the time-domain cannot characterize the states of mind well, the data has been computed in the frequency-domain using wavelet and Fourier transform. After the wavelet decomposing by db4 wavelet function, eight eigenvalues have been distilled and these eight eigenvalues are: EMG's σ_m , ca5_e, cd1_e;HR'sah, ca5_e, cdl_e, rd2_range;respiration volume's rdl_range.According to the collection of eigenvalues corresponding to various states, fatigue state can be recognized in two steps by fuzzy C-Mean arithmetic. The first step is to classify the states into normal and abnormal state by MCis_e;then, partition the abnormal states into medium and extreme phases. In the second step, four fuzzy membership functions can be built by