【作者】 罗莉华；

【导师】 李平； 王慧；

【作者基本信息】 浙江大学 ， 控制科学与工程， 2011， 博士

【摘要】 汽车自适应巡航控制(ACC:Adaptive Cruise Control)系统作为一种先进的驾驶辅助系统,旨在减轻驾驶员的精神负担,减少因驾驶员失误而造成的交通事故、提高汽车乘坐舒适性和燃油经济性、并改善交通流等,因而近年来已得到了政府、企业以及高校研究机构的广泛关注。在汽车ACC系统的控制单元中,安全车间距策略以及系统的控制算法是实现ACC系统的功能及其实用化的关键所在,其设计的好坏直接决定了ACC系统的动态响应。而对由ACC汽车构成的交通流建立对应的宏观模型,不仅有利于分析ACC车流的动态演化规律、揭示ACC系统对交通流的改善作用,而且为交通管理者进行决策和控制提供理论依据。本文以汽车ACC系统为研究对象,分别对间距策略、针对车间相互纵向动力学特性的上层控制、结合车间纵向动力学和油门刹车切换特性的一体化控制、ACC车流的宏观交通流模型进行了研究。全文主要研究成果总结如下：(1)提出了一种考虑前车速度趋势的可变跟车时距策略,并对其间距误差的收敛稳定性进行了理论证明。该间距策略通过引入对前车未来速度扰动的考虑提高了间距控制的前瞻性以及抗干扰能力,并通过饱和函数的处理,使车头时距更为合理的同时改善交通流的通行能力。与现有车间距策略相比,该间距策略能够适应复杂多变的行驶环境,有效地平衡了行驶过程中的安全性和跟车性,改善了间距控制的动态性能。(2)在模型预测控制(MPC:Model Predictive Control)的框架下提出了一个兼顾安全性、跟车性、舒适性和燃油经济性的多目标ACC上层控制策略。首先建立了表征车间相互纵向动力学特性的高阶ACC上层控制模型；接着在该模型的基础上深入分析了车辆行驶的多个控制目的,并将其转化为相应的性能指标和系统约束,在MPC的框架下设计了一个多目标ACC上层控制策略。相比于现有的上层控制策略,该策略不仅保证了ACC车辆在行驶过程中的安全性和跟车性,而且显著改善了乘坐舒适性和燃油经济性。(3)在MPC的框架下提出了一种考虑驾驶员行驶特性的双模式多目标ACC上层控制策略。该策略以对NGSIM实测微观驾驶数据的分析为基础,由平稳跟车模式、快速接近模式、以及模糊切换策略组成。当两车间距处于期望值附近时,该双模式ACC策略采用平稳跟车模式,利用上述第二点提出的算法,安全、平稳地跟踪前车；一旦车间距远大于期望值时,该策略切换到快速接近模式,设计了一个时间最优的多目标MPC算法,将其转化为一个混合整数非线性规划(MINLP:Mixed Integer Nonlinear Programming)命题,并提出了一种基于粒子群(PSO:Particle Swarm Optimization)算法的双层嵌套算法对其进行有效求解。为了模拟了驾驶员的决策过程,基于模糊推理建立了不同模式间的切换逻辑。仿真结果表明：该双模式ACC上层控制策略在满足多个行驶目的的同时,有效地模拟了驾驶员的行驶特性,根据行驶环境选择不同的行车策略,避免了换道乱插队现象的发生,因而有利于提高ACC系统的使用率。(4)提出了一种结合车间相互纵向动力学特性和油门刹车机械特性的ACC一体化控制结构,并在该控制结构下利用MPC的框架设计了油门刹车优化切换的多目标控制策略。首先通过引入逻辑变量并加入逻辑不等式约束将油门、刹车的机械特性统一在一个模型框架内,继而结合车间纵向动力学特性得到了ACC的综合模型；接着在该模型基础上设计了油门刹车优化切换的一体化多目标控制算法,在满足道路行驶过程中的安全性、跟车性、舒适性、节油性等多个目标的同时,对车辆自身的油门刹车切换性能(切换序列、对应执行器输入量)进行优化,从而在MPC的框架下转化为一个MINLP命题,并采用上述第三点提出的基于PSO的双层嵌套算法对其进行有效求解。与经典的阈值切换策略相比,该策略在保证行驶过程中的多个目的的同时,有效减少了油门、刹车间的切换次数,避免了因不同执行机构频繁切换而带来的机械磨损,改善了执行器的动态响应。(5)建立了ACC汽车的宏观交通流模型。该模型结合了ACC车辆的控制模型以及交通流的基本特性,真实地反映了宏观ACC车流的稳态特性和动态演化规律,有效地揭示了交通流动力波的传动特性。通过分析该宏观模型的速-密特性和流-密特性,证明了ACC系统的使用能显著提高道路的通行能力和服务水平,从而有效改善交通流。更多还原

【Abstract】 As an advanced Driver Assistant System (DAS), Adaptive Cruise Control (ACC) system is designed to release human drivers’mental load, reduce the car accidents, increase the driving comfort, decrease the fuel consumption and improve the traffic flow. In recent years, ACC system has received considerable attention from government, automobile companies, and research institutions.In the control module of ACC system, spacing policy and the control algorithm are the key factors to achieve system function and its real application, so they directly determine the dynamical responses of ACC systems. Establishing the macroscopic traffic flow model of ACC-equipped vehicles is not only beneficial to analyze the macroscopic dynamical characteristics of ACC flow, reflect the improvement of ACC system to the traffic flow, but also provides theoretical support for traffic management and control. This dissertation takes ACC system as the research object, investigate the spacing policy, the upper level control algorithm which aims at the longitudinal dynamics of inter-vehicle, the integrated control design including the longitudinal dynamics of inter-vehicle and the switch performance of throttle and brake, and the macroscopic traffic flow model of ACC flow, respectively. The main contributions of this dissertation are presented as follows:(1) A variable time headway policy which considers the velocity tendency of preceding vehicle is proposed, and the corresponding stability of spacing error is proved. By introducing the acceleration disturbance of preceding vehicle, the ability regarding forward-looking and anti-disturbance of spacing control is improved. Moreover, the introduction of saturation function makes time headway more reasonable and meanwhile improves the capacity of traffic flow. Compared to the traditional spacing policies, the proposed one can adapt the complex traffic scenarios, effectively balance the safety and car-following, and improve the dynamical performance of spacing control.(2) A multi-objectives upper level controller of ACC is proposed in MPC (Model Predictive Control) framework, which can satisfy safety, car-following, driving comfort and fuel efficiency. First, a high order state model which describes the longitudinal dynamics of inter-vehicle is established. Based on this model, the control objectives of ACC system are analyzed and transformed to the performance index and system constraints, respectively. Then the control algorithm is designed in MPC framework to satisfy these multi-objectives and constraints. It is shown that the proposed ACC upper-level control algorithm not only meets the safety and car-following requirements, but also outperforms the traditional algorithms by improving driving comfort and reducing fuel consumption.(3) A two-mode upper level controller of ACC in MPC framework is designed based on humans’driving habits. By analyzing the real microscopic traffic data from NGSIM, the control algorithm is designed to consist of steady following mode, fast approaching mode and fuzzy switch logic. Steady following mode is activated when the inter-distance falls round the desired value. The control algorithm proposed in (2) is utilized for this mode, with the purpose of following the preceding vehicle safely and steadily. Once the inter-distance becomes larger than the desired value, ACC vehicle switches to the fast approaching mode, which employs a time-optimal MPC algorithm. Then the controller design is transformed to be an online Mixed Integer NonLinear Programming (MINLP), which is solved by a nested two-loop algorithm based on Particle Swarm Optimization (PSO). In order to imitate human drivers’ decision-making, the switching strategy between the two control modes is developed based on fuzzy inference It is shown that this two-mode upper level ACC algorithm can satisfy the multi-objectives of vehicle traveling, choose different control modes according to the current traffic condition, and therefore effectively reflect the humans’driving habits and increase the usage of ACC system.(4) An integrated control structure is proposed, which synthesizes the longitudinal dynamics of inter-vehicie with the mechanical characteristics of throttle and brake. Moreover, an integrated ACC algorithm with optimal switching between throttle and brake is designed in this control structure. First, the binary integer variables and logic constraints arc introduced to synthesize the dynamics of throttle and brake into one model framework, and then it is combined with the longitudinal dynamics of inter-vehicle, which leads to an integrated ACC model. Based on this integrated ACC model, the control algorithm is designed to satisfy not only safety, car-following, driving comfort and fuel efficiency, but also the optimal switching performance (the switching sequence and the control input of actuator) between throttle and brake. Therefore, the controller design is transformed to be an online MINLP, and it can be solved by the nested two-loop algorithm based on PSO, which is proposed in (3) mentioned above. Compared to the tradition threshold switching algorithm, this control algorithm can not only satisfy the multi-objectives of vehicle traveling, but also effectively reduce the switching between throttle and brake, avoid the mechanism abrasion caused by frequent switching between different actuators, and improve the dynamics of control input of actuators.(5) A macroscopic traffic flow model of ACC-equiiped vehicles is established. By synthesizing the ACC control algorithm and the principles of traffic flow, this macroscopic ACC model truly reflect the stability features and dynamical evolution laws of ACC flow, and effectively reveal the transmission characteristics of traffic wave. Based on this macroscopic ACC model, the characteristics of speed-density and flow-density are analyzed. It is proved that the usage of ACC system is beneficial to increase road capacity and service ability, therefore effectively improve traffic flow.更多还原