基于正则极限学习机的驾驶员疲劳状态分类方法

Driver Fatigue Classification Method Based on Regularized Extreme Learning Machine

为避免接触式疲劳检测方法给驾驶员带来干扰,解决单一信号源对于反映疲劳程度可靠性低的问题,实现对疲劳状态高精度、高速度的检测,提出一种基于正则极限学习机的驾驶员疲劳状态分类方法。该方法通过多普勒雷达模块采集驾驶员生理信号,包括呼吸信号和心跳信号,作为神经网络输入数据。通过多源信息结合的方式提高疲劳状态检测可靠性。设计正则极限学习机(RELM)模型对数据集进行训练。实验结果显示,基于RELM算法模型检测驾驶员疲劳状态的准确率达92%。RELM算法可实现对训练数据的快速计算和学习,同时通过特征变换消除个体差异,实现对驾驶员疲劳状态较高的检测率。

In order to effectively solve the problems of interference caused by the contact fatigue detection method to the driver and the low reliability of a single signal source to reflect the fatigue level,and to achieve high-precision and high-speed detection of fatigue status,this paper proposes a driver fatigue classification method based on regularized extreme learning machine.This method uses the Doppler radar module to collect the driver’s physiological signals,including breathing signals and heartbeat signals,as input data of the neural network.In this paper,the reliability of fatigue state detection is improved by combining multiple sources of information.A regular extreme learning machine(RELM)is designed to train the dataset.The experimental results show that the accuracy of detecting the fatigue state of the driver based on the RELM algorithm model reaches 92%.The RELM algorithm can realize fast calculation and learning of training data,and at the same time,the method of eliminating individual differences through feature transformation can achieve a higher detection rate of driver fatigue.