面向计算流体力学的图形处理器资源管理

针对求解计算流体力学过程中图形处理器资源利用率低的问题,提出面向计算流体力学的图形处理器资源优化管理方案。基于计算流体力学的算法特性和同时运行任务的执行特点,设计合理的调度方案。通过动态改变不同任务的启动规模和启动时间,在减少资源竞争的同时提高图形处理器资源的有效使用。实验结果表明:本文提出的资源管理方案相比基线方法在不同任务规模下的平均加速比达到 1.64,对图形处理器的硬件资源使用也有了显著的提升。     

Efficient algorithms for flexible job shop scheduling with parallel machines

Job shop scheduling with a bank of machines in parallel is important from both theoretical and practical points of view. Herein we focus on the scheduling problem of minimizing the makespan in a flexible two-center job shop. The first center consists of one machine and the second has k parallel machines. An easy-to-perform approximate algorithm for minimizing the makespan with one-unit-time operations in the first center and k-unit-time operations in the second center is proposed. The algorithm has the absolute worst-case error bound of k − 1 , and thus for k = 1 it is optimal. Importantly, it runs in linear time and its error bound is independent of the number of jobs to be processed. Moreover, the algorithm can be modified to give an optimal schedule for k = 2 .     

雷达装备战场抢修支援分队调度决策研究

针对雷达装备战场抢修支援分队支援抢修多个现场的损伤雷达装备,以及抢修过程中可能出现其它现场损伤雷达装备请求支援抢修的调度问题,以实现每次支援抢修损伤雷达装备都获得最大的战斗力恢复率为目标,提出了一种类似于计算机中断过程的调度决策方法。调度决策中的抢修中断优先级判定方法,也可作为抢修支援分队确定支援抢修多个现场损伤雷达装备初始次序的依据。最后,通过实例应用验证了调度决策方法的实用性和可行性。     

电子侦察卫星联合侦察的混合调度算法

针对电子侦察卫星的使用约束,及不同任务的调度需求,建立了电子侦察卫星联合侦察的多目标混合整数规划模型.利用进化算法的全局搜索能力和变邻域搜索的局部优化能力,提出了一种多目标进化算法和变邻域搜索相结合两阶段混合调度算法MOEA VNS.针对问题多时间窗组合优化特点,设计了进化算子与邻域移动算子,在确保解多样性的同时使算法...     

Bounded flexibility in days‐on and days‐off scheduling

This article presents a flexible days‐on and days‐off scheduling problem and develops an exact branch and price (B&P) algorithm to find solutions. The main objective is to minimize the size of the total workforce required to cover time‐varying demand over a planning horizon that may extend up to 12 weeks. A new aspect of the problem is the general restriction that the number of consecutive days on and the number of consecutive days off must each fall within a predefined range. Moreover, the total assignment of working days in the planning horizon cannot exceed some maximum value. In the B&P framework, the master problem is stated as a set covering‐type problem whose columns are generated iteratively by solving one of three different subproblems. The first is an implicit model, the second is a resource constrained shortest path problem, and the third is a dynamic program. Computational experiments using both real‐word and randomly generated data show that workforce reductions up to 66% are possible with highly flexible days‐on and days‐off patterns. When evaluating the performance of the three subproblems, it was found that each yielded equivalent solutions but the dynamic program proved to be significantly more efficient. © 2013 Wiley Periodicals, Inc. Naval Research Logistics 60: 678–701, 2013     

基于遗传算法的多功能电子系统任务调度

雷达、通信、电子战等多功能一体化是未来作战平台电子系统的发展方向之一,任务优化调度技术是多功能电子系统的关键技术。针对多功能电子系统射频任务调度问题,在建立任务调度数学模型的基础上,采用了改进的遗传算法对模型进行了求解,提出了任务列表和随机键结合的编码方式,设计了以执行时间最短和任务丢失率最小为目标的多目标适应度函数,采用轮盘赌选择、改进型双点交叉和对称交换变异方式设计遗传操作。通过仿真实验,验证了所设计任务调度方法的有效性和优越性。     

双舰编队协同制导防空作战的动态拦截排序问题

针对双舰编队协同制导作战条件下如何对空袭目标进行排序这一问题,依据作战时间给出了排序算法。首先对平台的作战时间进行分析和求解,将射击周期分成了2段,把静态拦截排序问题转化为了2台处理机的同顺序作业排序问题,然后根据输入信息的改变而不断调整空袭目标的顺序以解决动态拦截排序问题,最后在想定条件下仿真计算,结果表明该算法具有较好的适用性。     

SpaceFibre星载数据网络低延时确定性调度算法

针对超高速SpaceFibre星载网络中多源数据传输的确定性和实时性应用需求,提出一种分类细粒度低延时确定性调度算法。该算法基于差异化调度策略的思想,将数据流划分为三类。为实现网络资源的细粒度分配,引入扩展时隙。该算法采用无冲突均匀调度方法,降低了数据包的平均排队延时。为适应有效载荷组网的航天应用场景,该算法兼顾网络拓扑结构生成调度方案。为验证算法有效性,在OPNET仿真平台下利用自定义建模技术搭建网络仿真模型。仿真结果表明:相比优先权调度和无冲突连续调度机制,该算法实现了时间敏感数据流的确定性传输;随着时隙数目的增加,网络的延时性能和抗抖动性能显著提升,吞吐量性能得到保证;该算法具有一定的航天工程实用价值。     

Vehicle routing problems with regular objective functions on a path

We investigate the problem of scheduling a fleet of vehicles to visit the customers located on a path to minimize some regular function of the visiting times of the customers. For the single‐vehicle problem, we prove that it is pseudopolynomially solvable for any minsum objective and polynomially solvable for any minmax objective. Also, we establish the NP‐hardness of minimizing the weighted number of tardy customers and the total weighted tardiness, and present polynomial algorithms for their special cases with a common due date. For the multivehicle problem involving n customers, we show that an optimal solution can be found by solving or O(n) single‐vehicle problems. © 2013 Wiley Periodicals, Inc. Naval Research Logistics 61: 34–43, 2014     

Improved bounds for batch scheduling with nonidentical job sizes

Here, we revisit the bounded batch scheduling problem with nonidentical job sizes on single and parallel identical machines, with the objective of minimizing the makespan. For the single machine case, we present an algorithm which calls an online algorithm (chosen arbitrarily) for the one‐dimensional bin‐packing problem as a sub‐procedure, and prove that its worst‐case ratio is the same as the absolute performance ratio of . Hence, there exists an algorithm with worst‐case ratio , which is better than any known upper bound on this problem. For the parallel machines case, we prove that there does not exist any polynomial‐time algorithm with worst‐case ratio smaller than 2 unless P = NP, even if all jobs have unit processing time. Then we present an algorithm with worst‐case ratio arbitrarily close to 2. © 2014 Wiley Periodicals, Inc. Naval Research Logistics 61: 351–358, 2014