基于坐垫体压分布的驾驶员生理状态识别方法

Recognition of Drivers' Physiological States Based on Body-Seat Pressure Distribution

提出一种基于坐垫体压分布识别长途驾驶过程中驾驶员生理状态的方法。开展长途驾驶模拟试验,采集9名被试者在驾驶过程中的坐垫体压分布数据,提取82维分类特征,再采用序列浮动前向选择算法从中选出55维最优特征子空间。试验结果表明,该方法能准确识别驾驶员生理状态,贝叶斯分类正确率可达93.37%。

A method to recognize drivers＇ physiological states during long distance driving based on seat- cushion pressure distribution is proposed. A simulation test of long distance driving is conducted, the seat-cushion pressure distribution data of 9 drivers are collected, from which 82-D classification features are extracted and 55-D optimal feature subspaces are selected by using sequential floating forward selection （SFFS） algorithm. The results of test show that the method proposed can accurately recognize drivers＇ physiological states, achieving a correct clas- sification rate of 93.37% with Bayes classifier.