基于深度强化学习的车辆跟驰控制

针对自适应巡航控制系统在控制主车跟驰行驶中受前车运动状态的不确定性影响问题，在分析车辆运动特点的基础上，提出一种能够考虑前车运动随机性的跟驰控制策略。搭建驾驶人实车驾驶数据采集平台，招募驾驶人进行实车跟驰道路试验，建立驾驶人真实驾驶数据库。假设车辆未来时刻的加速度决策主要受前方目标车辆运动影响，建立基于双前车跟驰结构的主车纵向控制架构。将驾驶数据库中的驾驶数据分别视作前车和前前车运动变化历程，利用高斯过程算法建立了前车纵向加速度变化随机过程模型，实现对前方目标车运动状态分布的概率性建模。将车辆跟驰问题构建为一定奖励函数下的马尔可夫决策过程，引入深度强化学习研究主车跟驰控制问题。利用近端策略优化算法建立车辆跟驰控制策略，通过与前车运动随机过程模型进行交互式迭代学习，得到具有运动不确定性跟驰环境下的主车纵向控制策略，实现对车辆纵向控制的最优决策。最后基于真实驾驶数据，对控制策略进行测试。研究结果表明：该策略建立了车辆纵向控制与主车和双前车状态之间的映射关系，在迭代学习过程中对前车运动的随机性进行考虑，跟驰控制中不需要对前车运动进行额外的概率预测，能够以较低的计算量实现主车稳定跟随前车行驶。     

Efficient multiple model particle filtering for joint traffic state estimation and incident detection

This article proposes an efficient multiple model particle filter (EMMPF) to solve the problems of traffic state estimation and incident detection, which requires significantly less computation time compared to existing multiple model nonlinear filters. To incorporate the on ramps and off ramps on the highway, junction solvers for a traffic flow model with incident dynamics are developed. The effectiveness of the proposed EMMPF is assessed using a benchmark hybrid state estimation problem, and using synthetic traffic data generated by a micro-simulation software. Then, the traffic estimation framework is implemented using field data collected on Interstate 880 in California. The results show the EMMPF is capable of estimating the traffic state and detecting incidents and requires an order of magnitude less computation time compared to existing algorithms, especially when the hybrid system has a large number of rare models.     

新能源汽车电池管理方法综述

电池管理系统作为电动汽车的重要组成部分,其中准确估测荷电状态(SOC)和电池均衡管理是电池管理系统的核心功能,也是优化系统能量管理,提高动力电池的使用效率,延长电池使用寿命的关键,对电池的储能至关重要。详细介绍主要SOC的估测方法和电池均衡管理方法,并对其优劣性进行分析与阐述,通过文献综述与对比分析为电池组在使用时提出理论依据与维护措施,为未来的研究提供便捷研究方法。     

基于工作电压的动力电池剩余容量动态估算

以单体蓄电池为研究对象，为解决电动汽车动力电池在工作状态下的剩余容量（SOC）进行实时估算的难题，提出了一种新的动力电池剩余容量估算方法。通过外接一个处于工作状态下电池周期性的恒定电流负载，测得一系列的电池工作电压，这些工作电压采用阶段斜率的计算方法求电池该时刻的剩余时间，从而推算出电池的SOC。证明了动态工作点折算到静态工作点的可行性。该方法的相对误差最大不超过7％，通过软件可实施对温度、老化等因素的补偿。误差分析结果表明，该方法误差小，测量精度高。     

A Novel Approach to Vehicle Dynamics using the Theory of a Cosserat Point and its Application to Collision Analyses of Platooning Vehicles

A fully three-dimensional, computationally inexpensive vehicular model is presented. In contrast to traditional rigid body models, the vehicle's sprung mass is modeled as a (nonlinearly) deformable body. The formulation of the equations of motion is based on a continuum theory known as the theory of a Cosserat point. These equations largely preserve the relative simplicity of rigid body dynamics but incorporate deformable features. The ease of computer implementation permits the simultaneous simulation of vehicle and collision dynamics of multiple vehicles and highway objects. In this paper, the theory of a Cosserat point is summarized and its general application to vehicle and collision dynamics is illustrated. A three-dimensional collision algorithm is discussed with emphasis on small closing velocities (negligible residual crush, elastic rebound). The novel model is compared to standard procedures.     

To fully automate or not? Investigating demands and willingness to pay for autonomous vehicles based on automation levels

We review the literature on the motivators of consumers’ purchasing decisions regarding autonomous vehicles (AVs), focusing on environmental awareness, vehicle types, concerns about accidents, and merits. The willingness to buy (WTB) and willingness to pay (WTP) for AVs have been extensively studied. However, it is imperative to broaden the outlook and consolidate the existing knowledge base. Although the WTB and WTP for AVs have been studied, automation-level preferences have been underresearched. Based on more than 150,000 observations, we construct a choice model and calculate the WTP for each level of automation. Our results show that there is a disparity between people's WTB and WTP according to the automation level. In particular, we find that the coefficients regarding FAVs’ benefits and accidents significantly exceed those related to environmental concerns. Such results indicate that practical benefits and concerns influence AV demand and WTP to a greater extent than environmental awareness. Our model results indicate a disparity between WTB and WTP for AVs that mostly derives from the different types of environmental concerns, concerns about accidents, and perceived benefits from these automobiles. Our results urge policies according to which governments and companies closely examine consumers who exhibit WTB-WTP disparities. Issues relating to demand patterns, WTP, and a suitable policy framework are discussed.     

水声网络探测节点的一种运动目标参数估计算法

文章提出了一种水声网络探测节点的运动目标参数估计方法,即基于模型匹配原理的估计算法。给出了算法的原理和仿真分析。     

贺尔海枢纽立交方案比选

对贺尔海枢纽立交的方案比选设计进行了详细的介绍,希望能够为同类立交设计时提供技术借鉴。     

基于序贯代理模型的转子系统频响特性区间分析方法

在转子系统的频率响应分析中,不确定性广泛存在,并且对分析结果有较大的影响。本文提出了一种面向含区间不确定性转子系统频响分析的序贯代理模型方法。基于区间数学以及转子动力学理论,本文首先建立了含区间参数转子系统频响的控制方程。基于传统的静态代理模型提出了面向转子系统区间分析的序贯代理模型方法。该方法首先通过叫少量的样本点构造粗糙代理模型,再提出了两个指标函数进行下一阶段的选点。重复该过程,直到满足收敛条件,最终可以得到转子系统频率响应的精细代理模型,进而得到区间边界。和传统的区间分析方法相比,该方法精度高,适合非线性问题。     

三种基于四元数模型的声矢量阵MUSIC算法

传统的基于长矢量模型的MUSIC算法没有充分利用矢量水听器各分量之间的正交特性,存在多目标分辨能力不足的缺点。针对这一情况,文中引入了四元数的概念,利用四元数能更好地描述矢量水听器各分量的正交结构这一优点,提出了三种基于四元数模型的二维声矢量阵方位估计算法:Q-VV算法、Q-PV算法和Q-PVV算法。与基于长矢量模型的MUSIC算法相比,新算法降低了协方差矩阵的维数,占用更少的内存空间。同时,由于利用了四元数强正交性约束的特点,新算法有好的单目标估计精度和多目标分辨能力。其中:Q-PV算法和Q-PVV算法利用了声压与振速的相干性,抗各向同性干扰的能力优于Q-VV算法。仿真实验证明了算法的有效性。