| As China’s urbanization and increasing levels of motorization, the problem of road traffic safety is increasingly serious. Fatigue Driving is one of the important factors. Therefore, we should work out an effective supervision and alarming system. It has significant meaning on cutting accident rate and protecting people’s life and property. Doing simulated experiment is an important approach to analyze typical factors which impact fatigue driving. It plays a vital role in studying fatigue driving mechanism and alarming method.In order to reduce the traffic damage caused by fatigue factors in driving, a method of driver fatigue detection based on EEG signal was proposed. As the traditional spectrum anal-ysis is not suitable for the character of transient signals, in the present study, it was aimed to decompose the EEG signal into frequency subbands using one-dimensional discrete wavelet transform to correspond to the EEG wave δ(1～4Hz),θ(4～8Hz), α(8～13Hz),β(13～30Hz) and a set of statistical features was extracted from the subbands to represent the distribution of wavelet coefficients as the inputs of the network. Then the state between sober and drowsy level was classified by using neural network with Levenberg-Marquardt algorithm. A part of the simulated drive data samples was chose to be trained to predict the other data. The result indicated that the method had a fairly good recognition and classification rate. The effect of parietal cerebral cortex on the sober state is better, while the effect of frontal cortex area on the drowsy better. |
