基于驾驶员模型参数辨识的疲劳驾驶研究

提出一种利用驾驶员模型反演方法来进行驾驶员疲劳诊断研究的新方法。首先利用预瞄神经网络建立适应于复杂路况条件下的驾驶员-汽车-道路闭环模型,然后定义特定行驶轨迹内理论数据与试验数据的近似度为目标函数,将驾驶员参数的反演问题转化为多目标优化问题,采用基于实数编码混沌变异量子遗传算法的优化方法,获得全局最优解。试验中采用脑电和主观疲劳心理评测结合的方法确定被试者的疲劳状况。在每种疲劳状况下对驾驶员参数进行辨识,对结果进行统计分析表明,在考虑到车型、道路曲率等因素条件下驾驶员参数分布与驾驶员的疲劳状况有很强的相关性。

Study on Fatigue Driving Based on Driver Model Parameter Identification

A novel research approach of fatigue driving based on the model parameter of driver quantum genetic inversion algorithm was proposed to improve the understanding of fatigue driving,which contains two steps.Firstly the preview neural network models are adopted to build a driver-car-road closed-loop model under complex road conditions;then,the degree of approximation between theoretical data and experimental data in special driving track is defined as the objective functions,thus the inversion problem of driver parameter is transformed into a multi-objective optimization problem.Lastly,a real-coded chaotic quantum-inspired genetic algorithm based on the objective functions is employed to obtain the global optimal solution for the model parameter of driver inversion.During the experiment,Examinee's fatigue state is measured by the method integrating EEG(Electro-encephalo-graph) evaluation with subjective psychological evaluation.According to the result of the identification statistic analysis,the parameter distributions are extremely correlated with the state of driver fatigue when considering complex conditions such as vehicle type and road curvature.