并行离散事件仿真中的DDM机制实现

在并行离散事件仿真系统中实现仿真模型间的互操作机制需要引入类似HLA中定义的声明管理(DM)和数据分发管理(DDM)服务.我们开发的KD-PADSE高性能并行离散事件仿真环境采用谓词断言和对象代理方法在面向对象的并行离散事件仿真系统中实现了基于平行结构和事件调度机制的声明管理(DM)和数据分发管理(DDM)服务.在以千兆以太网互联的10节点Linux集群计算机上,采用DDM机制后仿真模型每秒(物理时间)可以进行1万次属性更新,是没有采用DDM机制时的20倍,在功能性和性能上验证了该DDM机制在并行离散事件仿真系统中的适用性.     

基于并行模拟的多核集群系统性能预测和分析

针对多核集群系统所表现出的新的性能特征,提出了面向多核集群系统消息传递应用程序的并行模拟模型并设计、实现了一个并行模拟器MCPSim(Multi-core Cluster Parallel Simulator),MCPSim在功能模型和性能模型上体现了片内核间、结点内片间以及结点间等三个层次上消息通信的特点,同时支持对应用的消息数量、通信量等的百分比分布的profiling功能,采用PRIMEJ、acobi3D、NPB IS以及HPL等Benchmark程序对MCPSim进行了测试,结果表明MCPSim性能预测的精度优于BigSim,同时能够广泛应用于针对多核集群系统消息传递应用程序的性能分析中。     

Exact algorithms for scheduling multiple families of jobs on parallel machines

In many practical manufacturing environments, jobs to be processed can be divided into different families such that a setup is required whenever there is a switch from processing a job of one family to another job of a different family. The time for setup could be sequence independent or sequence dependent. We consider two particular scheduling problems relevant to such situations. In both problems, we are given a set of jobs to be processed on a set of identical parallel machines. The objective of the first problem is to minimize total weighted completion time of jobs, and that of the second problem is to minimize weighted number of tardy jobs. We propose column generation based branch and bound exact solution algorithms for the problems. Computational experiments show that the algorithms are capable of solving both problems of medium size to optimality within reasonable computational time. © 2003 Wiley Periodicals, Inc. Naval Research Logistics 50: 823–840, 2003.     

大容量储能发电机并联运行励磁控制算法

大容量储能发电机作为脉冲电源，整流后通过逆变器向大功率脉冲负载供电。采用储能发电机直流侧并联方式工作，可以有效提高系统带载能力与可靠性。当并联运行的储能发电机转速不同时，根据转速合理分配电机释放功率可以有效提高系统工作效能，因此需要设计一种具有功率调节功能的励磁控制算法。在电压、电流双闭环控制的基础上，引入功率前馈控制，推导并求解了前馈功率与励磁电流关系以及控制参数，并根据电机初始转速设计了功率分配算法。仿真试验结果表明，功率前馈控制在励磁控制过程中起主导作用，有效分配储能电机释放功率大小，提高了励磁电流响应速度，抑制了负载功率扰动对励磁控制系统的影响。     

装备剩余寿命预测平行仿真中模型动态演化方法

装备平行仿真中的一个重要概念是实时数据驱动的模型动态演化,但是至今仍缺乏具体应用领域的实现方法。以带未知离散冲击的混合退化装备剩余寿命预测为背景,以多态Wiener状态空间模型为演化对象,提出一种装备平行仿真中模型动态演化方法,包括基于交互多模型强跟踪滤波的模型软切换和基于期望最大化算法的模型参数在线估计,并实现了基于平行仿真的装备剩余寿命实时预测。利用某轴承退化数据进行实例研究,结果表明该方法能有效提高仿真逼真度,剩余寿命预测的准确度较高、不确定性较小,具有较高工程应用价值。     

地籍数据库点线拓扑一致性并行检查方法

针对拓扑检查算法复杂、计算量大,串行计算已远不能满足海量地籍数据高效拓扑检查需求的问题,在分析了点线拓扑关系的并行特点基础上,将界址点的数据划分方法与界址线的QR空间索引方法相结合,实现了界址点与界址线的并行拓扑计算。用某地区实际的界址点集与界址线集对点线拓扑并行检查进行实验。测试结果表明:并行检查算法的并行效率随着进程数的增加而有所衰减,但稳定在30%以上,加速比达到5以上,且相比于Arc GIS效率提升了30倍以上。并行检查方法以工具的方式集成应用于高性能地理计算平台中,应用效果良好。     

一种面向移动Agent的多任务并行计算模型及算法

介绍了一种面向移动Agent的并行计算模型,给出了采用十标度策略解决任务排序,采用满射策略解决任务映射的算法。该模型允许多个计算任务在异构主机构成的分布式环境下同时进行计算,并且通过算法优化,降低移动Agent之间的通信成本,减少网络流量。     

Parallel machine scheduling with eligibility constraints: A composite dispatching rule to minimize total weighted tardiness

We study a parallel machine scheduling problem, where a job j can only be processed on a specific subset of machines M_j, and the M_j subsets of the n jobs are nested. We develop a two‐phase heuristic for minimizing the total weighted tardiness subject to the machine eligibility constraints. In the first phase, we compute the factors and statistics that characterize a problem instance. In the second phase, we propose a new composite dispatching rule, the Apparent Tardiness Cost with Flexibility considerations (ATCF) rule, which is governed by several scaling parameters of which the values are determined by the factors obtained in the first phase. The ATCF rule is a generalization of the well‐known ATC rule which is very widely used in practice. We further discuss how to improve the dispatching rule using some simple but powerful properties without requiring additional computation time, and the improvement is quite satisfactory. We apply the Sequential Uniform Design Method to design our experiments and conduct an extensive computational study, and we perform tests on the performance of the ATCF rule using a real data set from a large hospital in China. We further compare its performance with that of the classical ATC rule. We also compare the schedules improved by the ATCF rule with what we believe are Near Optimal schedules generated by a general search procedure. The computational results show that especially with a low due date tightness, the ATCF rule performs significantly better than the well‐known ATC rule generating much improved schedules that are close to the Near Optimal schedules. © 2017 Wiley Periodicals, Inc. Naval Research Logistics 64: 249–267, 2017     

Rényi信息增量和协方差联合控制的多传感器管理算法

针对目标跟踪中雷达组网场景下多传感器管理问题,结合Rényi信息增量和协方差两种算法各自特性,利用并行处理的思想提出了一种基于Rényi信息增量和协方差联合控制的传感器管理算法。在具体仿真设计环节,分为传感器跟踪能力大于目标数和传感器跟踪能力小于目标数两种场景。仿真结果表明该算法在单目标匀速、多目标匀加速等多数场景下能够对目标进行有效跟踪,同时降低了传感器的切换频率,具有更好的实时性。     

基于栅格分层的逐栅格汇流算法并行化研究

分布式水文模型中的逐栅格汇流算法计算量大,需要借助并行计算以满足大流域长历时模拟的要求。针对目前鲜有对基于隐式有限差分的逐栅格汇流算法进行并行计算研究的情况,基于栅格分层的思想提出一种适用于共享内存并行计算环境的逐栅格汇流并行算法。该算法首先根据流向进行栅格分层,使同一层中栅格的计算相互独立,然后将同一层中栅格的计算任务分配到多个计算单元并行计算。采用C++编程语言与OpenMP并行编程库实现了该算法,并选择河北省清水河流域为实验区,在不同数据规模(30m、90m、270m分辨率)、不同核数(2～20个)以及不同栅格分层方法的情况下对算法性能进行了测试。实验结果表明本文提出的并行算法具有较好的加速比和并行效率,且并行效率随数据规模的增大而增大。栅格分层方法对算法并行性能有明显影响,从上到下的分层方法比从下到上的方法具有更高的并行效率。