Two-stage stochastic bilevel programming over a transportation network

We consider a two-stage stochastic extension of the bilevel pricing model introduced by Labbé et al. (1998). In the first stage, the leader sets tariffs on a subset of arcs of a transportation network, with the aim of maximizing profits while, at the lower level, flows are assigned to cheapest paths of a multicommodity transportation network. In the second stage, the situation repeats itself under the constraint that tariffs should not differ too widely from those set at the first stage, a condition that frequently arises in practice. We analyze properties of the model, provide numerical illustrations, and open avenues for further research into this area.     

Computing 4D near-optimal trajectories for dynamic air traffic flow management with column generation and branch-and-price

Abstract

The current air traffic system faces recurrent saturation problems. Numerous studies are dedicated to this issue, including the present research on a new dynamic regulation filter holding frequent trajectory optimisations in a real-time sliding horizon loop process. We consider a trajectory optimisation problem arising in this context, where a feasible four-dimensional (4D) trajectory is to be built and assigned to each regulated flight to suppress sector overloads while minimising the cost of the chosen policy. We model this problem with a mixed integer linear programme and solve it with a branch-and-price approach. The pricing sub-problem looks for feasible trajectories in a dynamic three-dimensional (3D) network and is solved with a specific algorithm based on shortest path labelling algorithms and on dynamic programming. Each algorithm is tested on real-world data corresponding to a complete traffic day in the European air traffic system; experimental results, including computing times measurement, validate the solution process.     

Line haul interstation spacing for low cost feeder modes

The choice behaviour of low cost travel (LCT) modes is very sensitive to travel distance. A line haul system designed on the basis of current planning practice of locating widely spaced stations to cater auto and bus feeder modes with the primary objective of gaining travel speed is hostile to non‐motorized and low cost feeder modes. With the revival of interest in promoting the use of walk'n ride and bike'n ride modes, there is a need to develop an appropriate tool to examine the effect of their specific characteristics in establishing the number and location of stations.

A generic normative behavioural hybrid model for locating the cost minimizing number and location of stations is developed for an LCT‐fed line haul system. The model considers the system with many to many two dimensional line haul demand density function in which the density varies in both x‐ and y‐directions. The feeder mode choice behaviour is incorporated in the model by integrating probability‐access/egress distance function with the objective function. Explicit functional relationships among the parameters of these feeder modes such as modal share as a function of access/egress distance with the parameters of line haul systems are developed. Dynamic programming is used to minimize the system cost. The generic model is shown to collapse into several simplified models capable of yielding approximate solutions for several well known special cases. It has been shown that location of stations is sensitive to the through load on board as well as users’ cost that defines the choice behaviour at large. Numerical examples are presented to demonstrate the applicability of the model.     

基于双层规划的危险货物配送路径鲁棒优化

针对不确定环境下带时间窗的多配送中心危险货物配送路径优化问题, 提出一种含鲁棒控制参数的鲁棒优化方法; 综合考虑危险货物运输风险、运输费用和服务时间窗, 构建了危险货物配送路径多目标双层鲁棒优化模型, 上层模型追求运输风险和运输费用最小化, 下层模型采用用户均衡交通分配模型; 根据Bertsimas-Sim鲁棒优化理论, 对含有不确定参数的上层模型进行鲁棒对等转化; 联合增强型Pareto遗传算法和Frank-Wolfe算法构建了求解多目标双层鲁棒优化模型的混合算法, 采用3段式编码和解码方法、等位匹配交叉操作以及翻转变异等遗传操作方法求解上层模型, 采用Frank-Wolfe算法求解下层用户均衡模型; 以经典的Sioux-Falls交通网络为例, 对含有3个配送中心、7个需求点的危险货物配送路径优化问题进行案例分析, 以验证模型及其算法的合理性。研究结果表明: 当鲁棒控制参数分别为0、30和60时, 构建的混合算法能分别快速得到3、2和3组鲁棒最优解, 且所有解均为包含具体运输路段和发车时刻的配送方案, 而非配送顺序; 该混合算法与传统两阶段启发式算法相比, 运算时间能节省54.74%。可见, 该混合算法无论是在求解效率上, 还是在解的表达形式上均优于两阶段启发式算法, 能较好地完成不确定环境下危险货物配送路径多目标双层鲁棒优化任务。     

城市交通疏散救援的元胞自动机模型

分析了城市内部各向车流冲突造成的车辆延误对疏散与救援决策的影响, 将交叉口控制参数引入到现有的基于元胞传输模型的应急疏散救援仿真模型中, 建立灾时交叉口紧急控制下的应急疏散救援仿真模型。改进后的模型以规划周期内疏散与救援加权行程时间最小为目标, 并引入反流策略。仿真结果表明: 由于低优先级的车流G1和G2让行高优先级车流G3和G4而产生延误, 其平均行程时间分别增加了10.0、11.1 s, 符合城市内部疏散与救援的实际情况; 反流策略实施后, 受益的车流G2和G4行程时间分别下降了6.5、6.0 s。可见, 提出的模型及反流策略有效。     

新建船队的船队优化模型建立和比较

船队规划问题通常是通过建立连续变量线性规划模型或混合整数线性规划模型来解决,但往往会得到要求一艘船营运于多条航线的结果。而企业出于安全因素和便于管理的角度考虑,希望船队单船个体的航线能固定,这就需要采用整数规划模型来解决此类问题。针对新建船队的整体配置规划问题,在建立连续变量线性规划模型和混合整数线性规划模型的基础上,探索并建立线性整数规划模型和非线性整数规划模型并进行优化。通过对计算实例的分析比较,论述了4种优化数学模型的特点与适用范围。     

数控加工技术浅析

数控加工技术已广泛应用于制造业中,努力发展数控加工技术,并向更高层次的自动化、柔性化、敏捷化、网络化和数字化制造方向推进,是当前机械制造业发展的方向.     

C语言在嵌入式控制系统开发中的应用研究

本文分析了嵌入式控制系统的特点和嵌入式控制系统开发的过程。用C语言开发嵌入式控制系统较汇编语言开发具有优势,结合实际的开发经验,重点分析了嵌入式C语言开发中应注意的问题以及代码优化问题。     

灾后道路抢修和物资配送的整合优化算法

为了在有限的时间内同时获得最佳抢修效益和物资配送方案, 提高救灾工作效率, 针对灾后道路抢修与救灾物资配送问题, 利用时空网络流动技巧, 构建了两者相应的时空网络, 在考虑两者的相关性后, 建立了多目标的灾后道路抢修工程与紧急物资配送混合整数多重网络规划模型, 提出了分解启发式求解方法。算例计算结果表明, 用CPLEX数学规划软件直接求解, 在求解到106.9 h时, 才可求得最优解, 而分解启发式方法只需31.8 min即可求得最优解, 其求解效率大幅提高, 求解时间对于实际的救灾工作是可以接受的。     

用西门子可编程控制器对T68镗床进行线路改造

用西门子可编程序控制器对T68镗床进行电器线路改造.机床原有的操作方式不变,机床的主电路不变,从而使机床的控制线路简化了,机床故障率降低了.文中首先提出电路改进的总体构思,设计出了PLC的梯形图及接线图,重点分析了部分电路的调试过程.