基于心率变异性的高速列车乘员舒适度评价研究

Research on comfort evaluation of high-speed train passengers based on heart rate variability

“以人为本”是我国高速铁路事业发展的基本出发点,高速列车乘坐舒适度的评估及优化是高速铁路技术发展的关键。而生理信号作为人体状态的直观反映,在评价乘员舒适度方面具有不可比拟的优势。心率变异性是心电信号中相邻心跳间期时间间隔的变化,可反映自主神经系统平衡状态。为研究心率变异性与高速列车乘员舒适度之间的关系,在沪昆高速铁路线上开展高速列车乘员舒适度实车实验,结合乘员主观舒适度问卷反馈结果,分析乘员的平均心跳间期、心跳间期标准差等10种心率变异性特征在舒适度劣化时的变化情况。对乘员综合舒适度、振动舒适度和耳气压舒适度3类不同问题,使用T检验法分析各特征对“舒适”和“不舒适”状态的显著性,并建立基于随机森林算法的高速列车乘员舒适度评价模型。研究结果表明,心率变异性特征在不同类问题舒适度状态改变时,基本具有一致的变化趋势;其中平均心跳间期(Mean_rr)与综合舒适度显著相关,心跳间期标准差(SDNN)及庞加莱图标准差(SD2)与振动舒适度显著相关;建立的舒适度识别模型对综合舒适度、振动舒适度及耳气压舒适度识别准确度分别达74.32%,70.67%及68.27%。

People-oriented development is the foundation of high-speed railway technology, making the evaluation of passenger comfort a key factor. Providing an intuitive reflection of the human body state,physiological signals has incomparable advantages in evaluating passenger comfort. Heart rate variability is the change of time interval between adjacent heartbeat intervals in electrocardiogram, which can reflect the balance state of the autonomic nervous system. To study the relationship between heart rate variability and passenger comfort of high-speed train, this paper presented a real vehicle experiment of passenger comfort of high-speed train on Shanghai-Kunming high-speed railway line. Combined with the results of the passenger subjective comfort questionnaire, the changes of ten heart rate variability characteristics of passengers under varying comfort levels were analyzed. To assess the comprehensive comfort, vibration comfort and ear pressure comfort of passengers, T-test was used to determine the significance of various characteristics on “comfortable” and “uncomfortable” states. Further, an evaluation model of passenger comfort of high-speed train based on random forest algorithm was established. The results show that the characteristics of heart rate variability have a similar change trend when the comfort level of different problems changes, and that Mean_rr is significantly correlated with comprehensive comfort, while SDNN and SD2 are significantly correlated with vibration comfort. The accuracy of the comfort recognition model for comprehensive comfort, vibration comfort and ear pressure comfort is 74.32%, 70.67% and 68.27%, respectively.