基于驾驶员脑电信号警觉度检测的车辆速度控制研究

发布时间：2018-08-17 10:37

【摘要】：随着疲劳驾驶引发的交通事故越来越多,以驾驶员为中心的汽车安全问题已经成为影响我国乃至世界道路交通安全的重要问题。开展以驾驶员为中心的汽车主动安全技术已经成为汽车行业一个研究热点,驾驶员警觉度检测设备将会逐渐成为高级驾驶辅助系统的一部分。故论文展开了对以驾驶员为中心的汽车主动安全技术的研究,提出了一种基于驾驶员脑电信号警觉度检测的车辆速度控制方法。学位论文的主要内容和研究工作如下：首先,完成无线可穿戴脑电信号采集器的制作。分析驾驶员脑电信号与疲劳之间的关系,包括脑电信号节律、不同脑区、不同脑电特征与疲劳之间的关系。在此基础上选择合适的脑区,传感器和信号处理设备,完成无线可穿戴脑电信号采集器的制作。然后设计实验完成对驾驶员处于清醒和疲劳状态下的脑电信号的采集。其次,分析脑电信号的干扰类型,采用小波滤波、去噪算法对脑电信号进行去噪预处理。提取脑电信号功率谱密度作为脑电特征用于脑电信号的分类。分析不同数据长度对于脑电特征的影响。而后,提出基于低秩矩阵分解技术的字典学习方法,将原始字典与训练数据集中的元素依次组成新的字典,进行低秩矩阵分解,然后提取出原始字典对应的低秩矩阵作为新的字典,迭代完成之后得到能够表示该类脑电信号的字典。提高了字典对元素的表示能力,减少了对训练数据的敏感性。然后再使用低秩矩阵分解技术进行警觉度的分类。最后,提出基于警觉度的车辆速度控制策略,用于决定什么时候、怎样对车辆进行速度控制(减速、刹车等)。提出基于车辆最小安全距离模型的车辆安全减速／刹车模型,用于计算车辆在速度变化过程中的加速度,防止在对车辆进行速度控制的时候发生追尾事故。最后建立车辆动力学模型对上述方法进行了验证,分析了不同情况下车辆速度变化曲线,得出本文所提出的基于驾驶员脑电信号警觉度检测的车辆速度控制系统有效率性结论。在论文最后,总结了现有成果,并进行下一步工作的讨论。
[Abstract]:With the increasing number of traffic accidents caused by fatigue driving, driver-centered vehicle safety has become an important issue affecting road traffic safety in China and the world. Driver-centered active safety technology has become a hot topic in automobile industry. Driver alertness detection equipment will gradually become a part of advanced driving assistant system. In this paper, the driver centered active safety technology is studied, and a vehicle speed control method based on driver's EEG alertness detection is proposed. The main contents and research work of the dissertation are as follows: firstly, the wireless wearable EEG collector is made. The relationship between EEG and fatigue in drivers was analyzed, including the rhythm of EEG, the relationship between EEG characteristics and fatigue in different brain regions. On this basis, the appropriate brain area, sensor and signal processing equipment are selected to complete the fabrication of the wireless wearable EEG collector. Then the experiment was designed to complete the EEG signal acquisition of the driver under the condition of waking and fatigue. Secondly, the interference type of EEG signal is analyzed. Wavelet filter and denoising algorithm are used to pre-process the EEG signal. The power spectral density of EEG is extracted as the feature of EEG to classify EEG signals. The effects of different length of data on EEG characteristics were analyzed. Then, a dictionary learning method based on low-rank matrix decomposition technique is proposed. The elements of the original dictionary and the training data set are composed of a new dictionary in turn, and the low-rank matrix is decomposed. Then the low-rank matrix corresponding to the original dictionary is extracted as a new dictionary and a dictionary capable of representing this kind of EEG signal is obtained after the iteration is completed. The ability of the dictionary to represent elements is improved, and the sensitivity of training data is reduced. Then the low rank matrix decomposition technique is used to classify the alertness. Finally, a vehicle speed control strategy based on alertness is proposed to determine when and how to control vehicle speed (deceleration, braking, etc.). A vehicle safety deceleration / braking model based on the vehicle minimum safe distance model is proposed to calculate the acceleration of the vehicle in the course of speed change and to prevent the rear-end accident when the vehicle is under speed control. Finally, the vehicle dynamics model is established to verify the above methods, and the vehicle speed change curves under different conditions are analyzed. The conclusion that the vehicle speed control system based on the driver's EEG alertness detection is efficient is obtained in this paper. At the end of the paper, the existing results are summarized, and the next work is discussed.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：U463.6

【相似文献】