汽车自适应巡航系统的建模及多目标优化切换控制策略研究

项目名称： 汽车自适应巡航系统的建模及多目标优化切换控制策略研究

项目编号： No.61304203

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 自动化技术、计算机技术

项目作者： 罗莉华

作者单位： 上海海事大学

项目金额： 22万元

中文摘要： 随着智能交通系统的推广，汽车自适应巡航控制（ACC）系统因能有效减轻驾驶员精神负担、减少交通事故、改善交通流等，已成为企业和学术界的研究热点。 本项目以ACC系统为研究对象，首先针对ACC车辆纵向动力学的混杂性和车间相互运动的不确定扰动影响，利用混合逻辑动态理论研究ACC系统的建模；其次，针对ACC系统的多目标控制需求，在预测控制框架下设计执行机构优化切换的多目标ACC控制策略，并研究多个目标优先关系的实现方法；最后，考虑到多目标ACC预测控制策略将转化为一个在线优化命题，本项目将从求解速度、收敛性、解的全局最优等方面综合考虑，设计求解算法。通过本项目研究，使ACC系统在保证行驶安全和高效跟车的基础上，通过减少执行机构的切换频率、优化执行器的控制输入特性和舒适度指标、平滑系统的动态响应，改善乘坐舒适度和燃油经济性。 研究成果可为开发具有自主产权的多目标ACC系统提供理论依据和控制算法基础。

中文关键词： 自适应巡航控制；混杂性；执行器优化切换；多目标；预测控制

英文摘要： With the spread of Intelligent Transportation System (ITS), vehicle Adaptive Cruise Control (ACC) system is becoming a hot research area in enterprises and academic field, since it can reduce drivers' mental load, decrease traffic accidents and improve traffic flow efficiently. In this project, ACC system is taken as the research object. Firstly, ACC system is modeled based on Mixed Logic Dynamic (MLD) theory, in order to describe the hybrid properties of vehicle longitudinal dynamics, and the uncertain disturbance influences in inter-vehicle kinematics. Secondly, a multi-objectives ACC control algorithm with optimal switching between throttle and brake is designed in Model Predictive Control (MPC) framework. Besides, how to formulate the priority relation among multiple objectives is investigated. Finally, the multi-objective MPC algorithm for ACC will be transformed to be an online optimization problem. Therefore, the solving algorithm is researched, with the consideration of solving speed, convergence, and the global optimum. According to this research, the ACC system not only meet the basic requirements of safety and efficient car-following, but also improve driving comfort and fuel-economy, by decreasing the switching frequency between throttle and brake actuators, optimizing the actuator inputs and comfo

英文关键词： adaptive cruise control；hybrid；optimal switching between throttle and brake；multi-objective；model predictive control