Study Of Driving Fatigue Based On Physiological Signal Collection

Driving fatigue has become one of the main factors in transportation. So, the driver fatigue detection and early warning have very important meaning and application value to avoid the occurrence of traffic accidents, to improve the working efficiency of the driver, and to protect the driver's health and safety.The thesis aims to reveal the relationship between the simulate cockpit ambient temperature, subjective evaluation, ECG parameters and the state of driving fatigue respectively. At the same time, it will take environmental factors and subjective evaluation results as parameters in the data analysis and fatigue state evaluation. A practical detection method of driving fatigue will be put forward at last. Firstly, the classification of driving fatigue detection methods seen in domestic and international material is introduced, especially those methods based on physiological signals. In order to have a preliminary understanding on factors which have obvious effects on driving fatigue, a basic information questionnaire is designed and distributed to 142 drivers. Based on data analysis of the feedback questionnaires, a driving fatigue simulation test is designed. Meanwhile, a data acquisition, storage and processing program is designed basing on the LABVIEW development environment, aiming for physiological signal acquisition. After a lot of driving simulation experiments under three typical ambient temperatures(15 degree, 25 degree, 35 degree), the subjective fatigue evaluation data and ECG data are analyzed. As a result, 8 kinds of ECG indexes(RR triangular index?TINN?VLF%?LF%?HF%?LF?HF and LF/HF) are found nearly irrelevant to ambient temperature. By subjective and objective contrast analysis and ROC curve analysis based on the index clustering value, initial thresholds of the 8 indexes are obtained. Then accuracy of the index initial thresholds is further analyzed, and 5 kinds of ECG indexes, HF%?LF/HF?RR triangular index?LF% and HF are found leading to higher fatigue detection accuracy.Research in the thesis provides a new thought for driving fatigue study. When the effects of ambient temperature need to be taken into consideration, such as in real world driving environment, ECG indexes which have little relation to ambient temperature are more suitable for driving fatigue detection. The initial driving fatigue index thresholds found in the thesis provide a good basis to detect driving fatigue and prevent the happening of driving fatigue in the real world, although they are preliminary and need further study.