Real-Time Detection and Early Warning of Driving Fatigue Based on Basic Scale Entropy

Abstract

Driving fatigue is the main cause of traffic accidents, and traffic accidents often cause serious personal injury and huge property losses. Therefore, it is of great significance for traffic safety to accurately and quickly detect the mental fatigue of the driver and perform fatigue warning. In this paper, preprocess the collected the electroencephalogram (EEG) signals to remove interference signals. The Butterworth band-pass filter is used to extract the EEG signals of α and β rhythms, and then the basic scale entropy of αand β rhythms is used as driving fatigue features. Using these features to analyze the driving fatigue state of the subjects in different driving stages, according to the change law of driving fatigue features and combined with the fatigue scale SOFI-25 (swedish occupational fatigue inventory-25), driving fatigue is divided into 3 levels (awake state, mild fatigue state and severe fatigue state). When the fatigue reaches a mild fatigue state or a severe fatigue state, a fatigue warning is given to the driver, and a piece of music that the driver is interested in is played. The results show that using the basic scale entropy of α and βrhythms as driving fatigue characteristics can effectively detect driver fatigue. The basic scale entropy is an entropy measurement algorithm with fast calculation, strong anti-interference and certain suppression of noise, which can realize real-time and more accurate detection of driving fatigue. In addition, when the fatigue reaches the state of mild fatigue or severe fatigue, this study provides fatigue warning to the driver, and plays a piece of music that the driver is interested in to relieve fatigue, which has practical value in actual driving and can effectively improve driving safety.