Having a full situational awareness while driving is one of the most important perceptions for safe driving which can be reduced by various factors such as in-vehicle infotainment, distraction, or mental load leading. Machine learning methods are being used to optimize for the identification of these inhibiting factors. To do so, three types of data were used: biographic features, physiological signals and vehicle information of 68 participants are being utilized to identify the normal and loaded behaviors. This research, therefore, concentrates on driving behavior analysis using a new automated hybrid framework for detection of performance degradation of drivers due to distraction. The proposed model contains a hybrid of extreme learning neural network, as an ensemble learning method and evolutionary algorithms, to determine the weights of classifiers, for combining several traditional classifiers. The obtained results showcase that the proposed model yields outstanding performance than the other applied methods.

驾驶时具有完整的态势感知是安全驾驶的最重要感知之一，可以通过各种因素来减少这种驾驶感，例如车载信息娱乐，分心或精神负担过大。机器学习方法正被用于优化识别这些抑制因素。为此，使用了三种类型的数据：68名参与者的传记特征，生理信号和车辆信息被用来识别正常行为和负重行为。因此，本研究着重于使用新的自动混合框架进行驾驶行为分析，以检测驾驶员因分心而导致的性能下降。提出的模型包含极限学习神经网络的混合体，作为整体学习方法和进化算法，确定分类器的权重，以结合多个传统分类器。获得的结果表明，所提出的模型比其他应用方法具有出色的性能。