【作者】 白子建；

【导师】 贺国光；

【作者基本信息】 天津大学 ， 管理科学与工程， 2007， 博士

【摘要】 目前,我国快速公交建设已经进入从导入期到成长期的发展阶段。快速公交系统规划、建设、运营管理逐渐得到了包括城市规划部门和公交运营部门在内的全社会范围的广泛关注。近年来新兴的智能优化算法的发展和应用,为解决快速公交系统的规划和发车频率优化提供了新的思路。本文在传统的普通公交系统的规划和发车频率优化的基础上,考虑到快速公交有别于普通公交的一些技术特性,应用智能优化新技术,针对快速公交系统的线网规划、战略层面的发车频率优化、战术层面的发车频率优化、战术层面的组合线路发车优化等问题进行了研究,概括起来,本文的工作主要包括以下几部分:1)考虑快速公交线网规划问题的特点,建立了一类直达乘客量最大化的BRT网络规划问题数学模型。并设计了求解该模型的禁忌算法,对算例进行仿真计算。该算法采用自然数编码,通过起、终点的配对及路线的生成以及路线车辆数的分配两个阶段构造初始解,设计了线路选取和车辆分配两种邻域分别对BRT线路以及发车频率进行操作。结合惩罚函数和邻域操作两种方法来处理各种约束。对有20个站点的BRT网络的线路规划进行了仿真试验。并通过调整算法的技术参数、确定参数重复计算等手段对仿真结果进行深入分析。2)针对快速公交客流在一个运营日中客流分布呈现“双驼峰”的特点,适时地调整分时段的发车频率,以保证运能的合理配置。在战略规划的层面上,主要从快速公交作为一个低收费的公共服务的角度分析问题,针对此类问题构建了兼顾公交公司和乘客双方利益的数学模型,并设计了求解该问题的遗传算法。遗传算法通过改进的交叉、变异操作设计等内容,对该类问题算例进行了仿真优化研究,同时还通过改变乘客等车时间权值,将优化结果与按客流比例发车情况下的结果进行对比分析等手段分析仿真影响因素和仿真效果。3)从快速公交公司作为一个独立运营经济实体角度出发,研究了公交运营规划的战术层面上发车频率优化设计问题,这样就考虑了快速公交与其他交通方式的客流竞争关系。根据票价、车辆速度和发车频率计算网络的交通广义出行费用,采用改进的logit分配法对路段的客流进行分配。对存在竞争关系的BRT发车频率优化问题构建了数学模型,设计了求解该模型的禁忌算法。禁忌算法采用自然数编码,把每天公交车的运行时间分成多个时间段。每一条公交线每时段的发车频率代表一个子解,所有子解组合成BRT发车频率优化问题的解。把违反站点最大通行能力约束和违反最大时间等待乘客比例约束作为惩罚项加入到目标函数中。采用发车频率之间的移动和时段发车频率独立移动两种邻域操作。并且通过算法灵敏度分析,优化前后的结果对比,问题灵敏度分析等内容对仿真结果进行深入研究。4)针对高峰时段经常出现的标准线路和大站快线组合发车的情况,考虑到与其他交通方式,主要是普通公交的客流竞争关系,从公交公司经营者的角度,在公交运营战术规划的层面上进行组合发车频率的优化。针对该类问题,建立了在保证一定服务水平的前提下追求公交公司经营利益的最大化的数学模型。结合禁忌算法和模拟退火算法二者的优点,提出了求解该模型的一种基于禁忌规则的模拟退火算法。在算法设计中,采用0-1编码表示快速线路站点集合,用自然数编码表示标准线路、快速线路的发车频率。根据BRT线路组合优化的特点,设计了单点取反、2-swap交换和单点增减三种邻域状态生成函数及相应的2-swap禁忌表和单点禁忌表,对组合线路算例进行优化仿真研究,并与单一算法的研究结果进行对比分析。更多还原

【Abstract】 Presently the Bus Rapid Transit in China has shifted into the growing-up stage from the beginning. The plan, construction and operation of BRT have been the focus of the many sectors including urban plan bureau and public transit corporation. In recent years, intelligent optimization algorithms have been newly developed and applied in solving the problems of BRT plan and frequency optimization. In this dissertation, the difference between BRT and traditional transit being considered, there are the researches on BRT network optimization, frequency optimization of stratagem, frequency optimization of tactics,combined routes optimization, etc.The main contents are as follows.1. A kind of BRT networks plan models for nonstop passenger flow maximization based on the problem characteristics is given. A tabu search algorithm as the solution is designed. The algorithm applies the coding by natural number to OD match, routes generation and vehicle numbers assignment to get original solution. And routes choosing and vehicle assignment respectively do with BRT routes and frequency with penalty function and neighborhood operation to resolve restrictions. In this part, a network planning with 20-stop BRT is simulated. Algorithm parameter regulation and repetition calculation are done for analyses.2. The BRT passenger flow distribution in one day is as Double-Peak, and the frequency in different periods must be adjusted for reasonable configuration of transportation capacity from the angel of stratagem planning, so a model considering the benefits of operator and passengers is built and the improved Genetic Algorithm for the problem is given. Simulation optimization based on the genetic algorithm is designed through betterment crossover and mutation operation. A simulation optimization research is given in which the effect of waiting time weight value to the result is as an analysis.3. From the angle of Transit Company as an economic entity, the competition between BRT and other transit mode is considered. Based on the calculation of cost function by ticket price, vehicle speed and frequency, an improved logit assignment method is applied to optimize the frequency and maximize the transit company benefit to keep the passengers’satisfaction service level. A mathematical model and solution algorithm based on tabu algorithm is designed for the transit dispatching with competition. In the tabu algorithm, nature coding is designed to divide one day into multiple periods so as to cut the solution set to parts according to the periods. And constrains of shift between different frequencies and unique shift are operated as two neighborhoods. Without change of fundamental parameters, vehicle speed and ticket price are changed to influence the simulation results.4. For the standard routes and pass-by bus combination, considering the competition from other modes, especially traditional transit, it is to optimize the frequency and maximize the transit company benefit for keeping the passengers’satisfation service level. Based on the characteristics of the problem, simulated annealing algorithm is combined with tabu search algorithm to optimize the frequency of BRT line combination. The combined tabu-simulated-annealing algorithm applies 0-1 coding to represent pass-by stops set and natural number to represent standard routes and frequency. According to BRT combined routes characteristic, the neighbor states of back single point, 2-swap exchange and add-cut single point are designed with 2-swap tabu table and single point tabu table. Then optimization simulation research is given for combined routes.更多还原