Psychological Stress Assessment Of Driver Based On Multi-physiological Parameters Analysis

With the rapid development of economy and the improvement of people’s living standards, the cars have gradually become the ordinary travel tools. Driving stress can induce boredom, fear, anger and so many different emotions, that directly affect the knowledge and decision-making abilities of the drivers and reduce the manipulation force,so that cause accidents on road. Therefore, the study of the characteristics of the drivers’ psychological stress can effectively alert the drivers, and reduce accidents, which has important significance for the normal operation of the transportation system.Physiological signals are a useful metric to provide a feedback about drivers’ physiological stress states, which can effectively identify the stress and emotion. This paper based on the multiple physiological signals to study the extent of drivers’ emotion under different levels of stress and achieve the assessment of psychological stress. The features of Electrocardio(ECG), respiration signals and electromyography(EMG) were extracted by the wavelet analysis, and 20, 18, 22 features about time domain and frequency domain respectively extracted. Again, the kernel principal component analysis was used to project the original feature sets onto a lower dimensional features space and retain the five eigenvalues to provide all the information, which is suitable for high-order problems. Finally, SVM can reflect differences in sample characteristics, and HMM can reflect the same sample similarity. Based on the analysis and comparison of stress recognition of SVM and HMM, the paper proposed an improved HMM based SVM to achieve stress assessment of the drivers. The results shew that maximum of recognition rate was 91.97% by SVM, maximum of recognition rate was 93.00% by HMM, maximum of recognition rate was 97.41% by the improved HMM based SVM.This model can assess the states of drivers’ psychological stress effectively, and feedback of the assessment will provide a guide to the drivers’ mental and physical interventions. Further, the driver is a type of high stress persons, building a assessment model for the specific population will play a supporting role for public traffic safety and management, but also for the assessment of the emotion states in other areas.