基于碰撞前车辆和驾驶员异常特征信号融合的主被动集成乘员防护方法研究

With the development of vehicle technology, the combination of active and passive technology of vehicle safety has become an important measure to improve the performance of occupant protection. However, the main present technologies are all based on expensive sensors and the high expense makes them difficult to promote. According to this research, there were always obvious abnormal vehicle and driver behaviors before crashes. Through the reasonable detection about the parameters of these abnormal behaviors, the technology can predict the crash based on low cost sensors. The technology can achieve a better performance of occupant protection by making full use of the margin of time and space before crashes. This research based on the simulation reconstruction and simulated driving analysis of classical crashes is to study the relationship between these abnormal behaviors and the crash before an accident happens and to extract the characteristic parameters for crash prediction. Through the combination of multiple signals, the research can establish a reliable crash prediction algorithm and enhance the performance of anti-interference. At the same time, with the seat belts pre-loading and active seat adjustment technology and the method of combination of computer simulation and experiment, the research can also study the optimal parameters of occupant restraint system before the crash under the condition of different figures and accidents. This research which has the advantages of low cost and wide application prospect can decrease the death rate of accidents effectively and has significant social and economic benefits.

随着汽车技术的不断发展，主被动集成汽车碰撞安全技术已成为提高乘员保护性能的重要措施。然而，目前主要的相关技术都是建立在昂贵的传感设备基础上，难以推广使用。本课题研究发现，汽车碰撞事故发生前通常具有明显的车辆和驾驶员行为异常特征，通过合理检测这些异常特征参数，可以采用低成本传感器实现对碰撞事故的预测。充分利用碰撞事故发生前的时间和空间裕度，可以达到更好的乘员保护效果。本课题拟通过对典型事故案例的仿真重建和模拟驾驶分析，研究碰撞事故发生前的汽车动态特征和驾驶员行为特征及其与碰撞事故的相关性，提取关键特征参数用于事故预测。通过对多种信号的融合建立可靠的碰撞预测算法，提高系统的抗干扰性能。结合安全带预紧及主动式座椅调节技术，利用计算机仿真与实验相结合的方法，研究针对不同身材和不同碰撞前乘员约束系统状态的最佳调节参数。该方法成本低，应用前景广泛，能够有效降低道路交通事故伤亡率，具有重大社会经济效益。

如果能在碰撞发生前就准确预测到危险性，将能够为汽车主被动安全系统提供足够的响应时间。基于雷达、视频等技术的碰撞预警系统已被广泛研究并在高档汽车中获得了应用，而降低成本将使碰撞预警系统获得更广泛的应用，从而进一步降低交通事故伤亡。本研究首先通过文献研究确定了车辆制动加速度和驾驶员的制动踏板角速度作为碰撞预警识别参数。然后通过大量的实车试验研究了车型、驾驶员、车速、制动强度等参数对碰撞预警识别参数的影响，制定了利用制动踏板角速度和车辆制动加速度进行碰撞预警的方法，并且指出制动踏板角速度具有更好地识别及时性。当使用移动窗积分方法进行信号处理时，推荐选择150ms作为移动积分窗宽，7.580×10-2rad作为紧急工况的识别阀值，此时紧急工况的诊断漏诊率和误判率分别为0和0.148。针对碰撞预警系统提出了可逆约束系统参数的匹配优化方法，将实验设计、Kriging代理模型以及第二代多目标遗传算法相结合，以乘员头、胸、颈、大腿的损伤值为优化目标进行多目标优化，并利用加权损伤指标WIC评价最优匹配方案，该方法能够获得良好的约束系统参数。针对主动预紧式安全带和主被动结合头枕开展了约束系统参数优化仿真及试验研究，提出了新型安全带和头枕的控制系统及结构方案并进行了样件试制，获得了良好的保护效果。同时对主动式安全座椅开展了仿真及试验研究，确定了优化方法并提出了最佳的主动式座椅匹配参数。另外，针对特殊体型乘员开展了基于有限元的约束系统优化方法研究，确定了针对不同体型的最优保护参数。目前，主动预紧式安全带和主被动结合头枕已与企业合作进行小批量试制，有望上市销售。