基于改进免疫遗传机理的动态路径诱导算法研究

针对传统算法存在的不足,提出基于改进的免疫遗传机理的动态路径诱导算法。通过引入单点交叉算子配合控制策略的改进思想,在宏观上控制交叉过程以及构造优化简单的目标函数,较好地弥补了传统遗传算法早熟收敛、陷入局部最优解、运算效率较低等问题。最后采用此算法进行路网实例计算,结果表明该算法能较大程度地提高路网利用率,减少交通拥堵。     

小件快运中的最短运输时间问题

小件快运经过多年的发展,已经成为公路客运一个新的经济增长点,越来越多的公路客运企业和汽车客运站加入到经营小件快运的行列,小件快运也逐渐由单个企业的独立运营发展到多个企业联合的网络化运营。对中转运输中多条运输线路进行分析,得到最短运输时间线路,从而提高运输效率。     

智能分布式驱动汽车路径跟踪/制动能量回收协同控制策略研究

为了解决智能分布式驱动汽车路径跟踪与制动能量回收系统间的协同控制难题,充分考虑分布式驱动汽车四轮扭矩独立可控在智能驾驶系统中的优势,设计适应不同路面附着条件的智能分布式驱动汽车转向、制动分层协同控制策略。上层控制器依据不同的路面类型设计差异化的多目标代价函数,以综合优化各工况下的控制目标。高附路面下,制定满足最大能量回收值的全局参考车速,在线优化路径跟踪指令,实现最优能量回收的同时减小系统运算负荷;低附路面下,优先考虑车辆的路径跟踪性能和行驶稳定性,在多目标代价函数中取消对全局参考车速的跟随要求,增设终端速度约束与能量回收项性能指标并减小能量回收项性能指标的权重系数。上层控制器基于模型预测控制方法对多目标代价函数进行滚动优化与预测求解,得到期望的前轮转角及4个车轮的总制动扭矩需求。下层控制器根据制动扭矩需求对四轮的液压制动扭矩和电机制动扭矩进行分配,最终完成整个复合制动过程。基于MATLAB/Simulink和CarSim软件,搭建控制器在环仿真平台,并在高附和低附路面条件下对所提出的策略进行试验验证。研究结果表明：高附路面下,所提出的控制策略在准确跟踪期望路径的同时相较固定比例制动力分配方法可提升2.7%的能量回收值并减少约0.02 s的单次计算时间;低附路面下,与使用高附控制策略相比,能够保证车辆的路径跟踪准确性与行驶稳定性,同时可提升7.8%的能量回收值;控制器在环试验结果证明了该协同控制策略对车辆性能提升的有效性。     

基于混合A*和可变半径RS曲线的自动泊车路径优化方法

路径规划是自动泊车系统的重要组成部分,是确保泊车运动安全、缩短行车距离、提高乘坐舒适性的关键。而当前自动泊车规划系统往往面临行驶空间狭小、障碍物多、路径搜索难度大等技术挑战,同时搜索曲线半径固定容易导致路径接点处曲率不连续,增大了路径跟随控制难度和轮胎磨损程度,这些都提升了泊车路径规划的研究难度。针对以上问题,设计可变半径的Reeds-Shepp曲线,提出基于混合A^*和该曲线的自动泊车路径规划方法,通过调整曲线半径,提升其在复杂场景下路径的搜索能力和灵活性。随后,设计基于分段贝塞尔曲线和梯度下降的路径优化方法,利用其多阶导数连续的优势优化已搜索的路径曲率,并采用梯度下降来保证路径曲率大小和对障碍的规避,解决直线与圆弧相接等位置曲率变化不连续的难题。结合路径搜索与路径优化的泊车规划方法能够切实满足复杂场景下的泊车需要。最后,基于团队自主研发的PanoSim虚拟系统与MATLAB搭建联合仿真环境,针对多种自动泊车工况测试验证提出的方法。研究结果表明：调整Reeds-Shepp曲线的搜索半径进行全局路径搜索,可获得更短和更易跟随的路径,具有良好的灵活性;基于贝塞尔曲线和梯度下降法的路径优化可有效消除曲率突变点、约束路径曲率并保证对障碍的无碰撞要求。     

公路隧道设计及施工技术问题分析

隧道作为组成公路整体结构的重要环节,其具体的设计及施工非常复杂,不断提高设计及施工技术应用的质量,才能够满足日益上涨的公路运输需求。从分析公路隧道建设项目中存在的设计及施工技术问题入手,提出设计和施工的优化方案,可以为提高公路建设质量提供有效的理论参考。     

Estimation of road profile variability from measured vehicle responses

When assessing the statistical variability of fatigue loads acting throughout the life of a vehicle, the question of the variability of road roughness naturally arises, as both quantities are strongly related. For car manufacturers, gathering information on the environment in which vehicles evolve is a long and costly but necessary process to adapt their products to durability requirements. In the present paper, a data processing algorithm is proposed in order to estimate the road profiles covered by a given vehicle, from the dynamic responses measured on this vehicle. The algorithm based on Kalman filtering theory aims at solving a so-called inverse problem, in a stochastic framework. It is validated using experimental data obtained from simulations and real measurements. The proposed method is subsequently applied to extract valuable statistical information on road roughness from an existing load characterisation campaign carried out by Renault within one of its markets.     

Trip generation and distribution: the inconsistency problem and a possible remedy

There are many shortcomings commonly associated with the conventional urban transportation modeling process. This paper focuses on one of the more important problems — the inconsistency between trip generation and distribution components — and suggests a possible way of alleviating it. The suggested approach involves sorting out the independent effects on tripmaking of origin, destination and travel cost characteristics, and introducing accessibility measures explicitly into the modeling process. The resulting modeling framework can be used to obtain consistent estimates of trip generation and distribution quantities which are responsive to changes in the transportation and spatial systems.     

Prioritizing new bicycle facilities to improve low-stress network connectivity

This paper introduces a new method to prioritize bicycle improvement projects based on accessibility to important destinations, such as grocery stores, banks, and restaurants. Central to the method is a new way to classify “bicycling stress” using marginal rates of substitution which are commonly developed through empirical behavioral research on bicyclist route choice. MRS values are input parameters representing bicycling stress associated with the number of lanes and speed limit of a street. The method was programmed as a geographic information system tool and requires commonly available data. The tool is demonstrated on three improvement scenarios that were recently proposed for Seattle, Washington. The full build-out scenario consists of 771 projects that include various new bike lanes, protected bike lanes, and multi-use trails. The tool produces priority rankings based on a project’s ability to improve low-stress connectivity between homes and important destinations. The analysis identifies specific areas and neighborhoods that can be expected to exhibit better bikeability. Transportation planners can use the tool to help communicate anticipated project impacts to decision-makers and the public.     

物流网络优化中路径问题的研究

在深入分析物流系统化和网络化的基础上,对物流网络优化中物流路径问题进行研究,通过分析物流结构模型和数学模型,提出了最短路径和关键路径物流策略的分析方法,给出了具体的计算方法,并举实例加以说明,为解决现代物流结构中存在的系统设计和运作等问题提供了理论依据和技术方法。     

混合不确定条件下绿色多式联运路径优化

主要研究了当运输时间、中转时间、客户需求和中转集拼货运量四重混合不确定因素服从随机分布时的绿色多式联运路径优化问题，运用随机优化理论，以运输成本、碳排放成本和时间惩罚成本为目标，建立混合不确定条件下绿色多式联运路径优化模型.通过对各子目标函数权重进行赋值，得出考虑不同成本因素的多式联运路径优化方案.探讨时间、需求和网络服务能力对多式联运路径优化结果的灵敏度分析，发现各成本随时间变动而变化的规律和边际运输成本最小时的服务时间；当货运量形成规模效应后可降低边际运输成本；不同网络服务规模的运输路径优化结果，以及满足客户不确定需求的最小网络配置.