基于非支配遗传算法的HLA仿真系统数据采集策略

doi:10.12305/j.issn.1001-506X.2024.09.22

摘要/Abstract

摘要：

数据采集是仿真执行过程中的重要环节, 数据采集的完整性和效率对整个训练仿真活动的最终效果和效率具有重大影响。然而, 在现有基于高层体系结构(high level architecture, HLA)的分布式仿真系统中, 集中式数据采集在单个步长内读写海量数据, 会影响仿真正常推进, 而分布式数据采集会造成大量冗余数据, 且采集模块的开发不具备通适性。针对上述问题, 基于弱分布式数据采集结构, 利用多个采集成员实现并行数据采集, 并基于非支配排序遗传算法Ⅱ(non-dominated sorting genetic algorithm Ⅱ, NSGA-Ⅱ)制定采集任务在多个成员间的分配策略, 实现数据采集负载的均衡分布。仿真结果和真实系统上的实验结果表明, 所提方法能显著提升数据采集效率, 同时减少数据采集成员执行过程中的中央处理器(central processing unit, CPU)和内存消耗。

关键词: 数据采集, 高层体系结构, 大规模分布式仿真, 非支配排序遗传算法Ⅱ, 采集效率

Abstract:

Data collection is the primary link in the simulation execution process. The integrity and efficiency of data acquisition have a significant impact on the final effect and efficiency of the entire training simulation activity. However, in the existing distributed simulation system based on high level architecture (HLA), centralized data collection is difficult to handle massive data in a single step, which will affect the normal advancement of simulation while distributed data collection will cause a large number of redundant data, and the development of the collection module does not have universality. In response to the above problems, based on weak distributed data acquisition structure, multiple collection members are used to realize parallel data collection, and the distribution strategy of collection tasks is formulated among multiple members by the non-dominated sorting genetic algorithm-Ⅱ (NSGA-Ⅱ) to achieve a balanced distribution of data collection loads. The experimental results on simulation result and real system show that the proposed method can significantly improve the efficiency of data collection while reducing the central processing unit (CPU) and memory consumption during the execution of data collection members.

Key words: data collection, high level architecture (HLA), large-scale distributed simulation, non-dominated sorting genetic algorithm-Ⅱ (NSGA-Ⅱ), collection efficiency

中图分类号:

TP391.9

王佩骐, 鞠儒生, 张淼, 段伟. 基于非支配遗传算法的HLA仿真系统数据采集策略[J]. 系统工程与电子技术, 2024, 46(9): 3103-3111.

Peiqi WANG, Rusheng JU, Miao ZHANG, Wei DUAN. Data collection strategy of HLA simulation system based on non-dominated genetic algorithm[J]. Systems Engineering and Electronics, 2024, 46(9): 3103-3111.

图/表 16

图1

图2

图3

表1

图4

表2

表3

表4

表5

表6

表7

表8

表9

图5

表10

图6

参考文献 34

1	HAN J W , KAMBR M . Data mining-concepts and techniques[M]. San Fransisco: Morgan Kaufmann, 2001.
2	FALCONE A , GARRO A . Distributed co-simulation of complex engineered systems by combining the high level architecture and functional mock-up interface[J]. Simulation Modelling Practice and Theory, 2019, 97, 101967. doi: 10.1016/j.simpat.2019.101967
3	IEEE 1516.1-2010 redline version, IEEE standard for modeling and simulation (M&S) high level architecture (HLA)-fede-rate interface specification-redline[S]. New York: IEEE, 2010.
4	MAGOUA J J , WANG F , LI N . High level architecture-based framework for modeling interdependent critical infrastructure systems[J]. Simulation Modelling Practice and Theory, 2022, 118, 102529. doi: 10.1016/j.simpat.2022.102529
5	宋恒杰, 杨明, 王子才. 基于HLA的仿真数据记录系统研究[J]. 系统工程与电子技术, 2005, 27 (3): 505-507, 544. doi: 10.3321/j.issn:1001-506X.2005.03.034
	SONG H J , YANG M , WANG Z C . Data collection system for HLA-based simulation[J]. Systems Engineering and Electronics, 2005, 27 (3): 505-507, 544. doi: 10.3321/j.issn:1001-506X.2005.03.034
6	吴跃平. 基于HLA的数据记录与回放系统的研究与实现[D]. 成都: 西南交通大学, 2012.
	WU Y P. The research and implementation of HLA-based data recording and playback system[D]. Chengdu: Southwest Jiaotong University, 2012.
7	鞠儒生, 乔海泉, 邱晓钢, 等. HLA仿真结果数据库设计及其应用研究[J]. 系统仿真学报, 2006, 18 (2): 327- 330. doi: 10.3969/j.issn.1004-731X.2006.02.016
	JU R S , QIAO H Q , QIU X G , et al. Research on design and application of HLA simulation result database[J]. Journal of System Simulation, 2006, 18 (2): 327- 330. doi: 10.3969/j.issn.1004-731X.2006.02.016
8	WANG W G , XU Y P , CHEN X , et al. High level architecture evolved modular federation object model[J]. Journal of Systems Engineering and Electronics, 2009, 20 (3): 625- 635.
9	张柯, 张新宇, 鞠儒生, 等. 基于HLA的分布仿真系统数据采集解决方案[J]. 系统仿真学报, 2004, 16 (12): 2725- 2728. doi: 10.3969/j.issn.1004-731X.2004.12.027
	ZHANG K , ZHANG X Y , JU R S , et al. Solution of data collection in HLA[J]. Journal of System Simulation, 2004, 16 (12): 2725- 2728. doi: 10.3969/j.issn.1004-731X.2004.12.027
10	刘庆国. HLA训练仿真系统数据记录与回放的设计和实现[J]. 火力与指挥控制, 2014, 39 (10): 121- 125. doi: 10.3969/j.issn.1002-0640.2014.10.031
	LIU Q G . The design and implementation of data recording and playback in a HLA-based tactical simulation system[J]. Fire Control and Command Control, 2014, 39 (10): 121- 125. doi: 10.3969/j.issn.1002-0640.2014.10.031
11	张柯, 张新宇, 鞠儒生, 等. 基于HLA的分布仿真系统数据采集解决方案[J]. 系统仿真学报, 2004, 16 (12): 2725- 2728. doi: 10.3969/j.issn.1004-731X.2004.12.027
	ZHANG K , ZHANG X Y , JU R S , et al. Solution of data collection in HLA[J]. Journal of System Simulation, 2004, 16 (12): 2725- 2728. doi: 10.3969/j.issn.1004-731X.2004.12.027
12	PAULINE A, WILCOX A G. BURGER P H. Investigating the impact of simulation communication patterns on distributed data collection strategies[C]//Proc. of the Spring Simulation Interoperability Workshop, 1999.
13	DROUGAS Y, REPANTIS T, KALOGERAKI V. Load ba-lancing techniques for distributed stream processing applications in overlay environments[C]//Proc. of the IEEE 19th International Symposium on Object and Component-Oriented Real-Time Distributed Computing, 2006.
14	TONG R X, ZHU X F. A load balancing strategy based on the combination of static and dynamic[C]//Proc. of the 2nd International Workshop on Database Technology and Applications, 2010.
15	TATBUL N, ZDONIK S. Dealing with overload in distributed stream processing systems[C]//Proc. of the 22nd International Conference on Data Engineering Workshops, 2006.
16	邓华锋, 刘云生, 肖迎元. 分布式数据流处理系统的动态负载平衡技术[J]. 计算机科学, 2007, 34 (7): 120- 123. doi: 10.3969/j.issn.1002-137X.2007.07.032
	DENG H F , LIU Y S , XIAO Y Y . Dynamic load balancing techniques for the distributed stream processing systems[J]. Computer Science, 2007, 34 (7): 120- 123. doi: 10.3969/j.issn.1002-137X.2007.07.032
17	GEDIK B . Partitioning functions for stateful data parallelism in stream processing[J]. The International Journal on Very Large Data Bases, 2014, 23 (4): 517- 539. doi: 10.1007/s00778-013-0335-9
18	唐颖峰, 陈世平. 分布式数据流处理系统管理中负载均衡问题建模与求解[J]. 运筹与管理, 2021, 30 (4): 155- 162.
	TANG Y F , CHEN S P . Modeling and solution for load balancing optimization in distributed stream processing system management[J]. Operations Research and Management Science, 2021, 30 (4): 155- 162.
19	ROBSON E , GRANDE D , BOUKERCHE A . dynamic balancing of communication and computation load for HLA-based simulations on large-scale distributed systems[J]. Journal of Parallel and Distributed Computing, 2011, 71 (1): 40- 52. doi: 10.1016/j.jpdc.2010.04.001
20	马武彬, 王锐, 王威超, 等. 基于进化多目标优化的微服务组合部署与调度策略[J]. 系统工程与电子技术, 2020, 42 (1): 90- 100. doi: 10.3969/j.issn.1001-506X.2020.01.13
	MA W B , WANG R , WANG W C , et al. Micro-service composition deployment and scheduling strategy based on evolutionary multi-objective optimization[J]. System Engineering and Electronics, 2020, 42 (1): 90- 100. doi: 10.3969/j.issn.1001-506X.2020.01.13
21	柳强, 毛莉. 基于NSGA-Ⅱ的油气管网多目标布局优化[J]. 控制工程, 2019, 26 (2): 308- 313.
	LIU Q , MAO L . Multi-objective layout optimization of oil-gas pipeline network based on NSGA-Ⅱ[J]. Control Engineering of China, 2019, 26 (2): 308- 313.
22	BHARGAV G , VIMAL S , VIVEK P . Multi-objective optimization of vehicle passive suspension system using NSGA-Ⅱ, SPEA2 and PESA-Ⅱ[J]. Procedia Technology, 2016, 23, 361- 368. doi: 10.1016/j.protcy.2016.03.038
23	陈通. 基于MOPSO算法的虚拟机动态整合策略研究[D]. 西安: 西安电子科技大学, 2021.
	CHEN T, Research of dynamic integration strategy of virtual machine based on MOPSO algorithm[D]. Xi'an: Xidian University, 2021.
24	AHMED B , OSSAMA M , ANAS S . A MOPSO algorithm based on Pareto dominance concept for comprehensive analysis of a conventional adsorption desiccant cooling system[J]. Journal of Building Engineering, 2022, 60, 105189. doi: 10.1016/j.jobe.2022.105189
25	何羚, 舒文江, 陈良. 改进的多目标粒子群优化算法及其在雷达布站中的应用[J]. 电子科技大学学报, 2020, 49 (6): 806- 811.
	HE L , SHU W J , CHEN L , et al. Improved multi-objective particle swarm optimization algorithm and its application in radar station distribution[J]. Journal of University of Electronic Science and Technology of China, 2020, 49 (6): 806- 811.
26	邵苗苗. 改进的MOEA/D算法在双目标投资组合问题中的应用[D]. 武汉: 华中师范大学, 2021.
	SHAO M M. An improved MOEA/D and application in bi-objective portfolio problem[D]. Wuhan: Central China Normal University, 2021.
27	FERNANDEZ E , RANGEL-VALDEZ N , CRUZ-REYES L , et al. Preference incorporation in MOEA/D using an outranking approach with imprecise model parameters[J]. Swarm and Evolutionary Computation, 2022, 72, 101097. doi: 10.1016/j.swevo.2022.101097
28	CAO J , ZHANG J L , ZHAO F Q , et al. A two-stage evolutionary strategy based MOEA/D to multi-objective problems[J]. Expert Systems with Applications, 2021, 185, 115654. doi: 10.1016/j.eswa.2021.115654
29	JIAO R W , ZENG S Y , LI C H , et al. Two-type weight adjustments in MOEA/D for highly constrained many-objective optimization[J]. Information Sciences, 2021, 578, 592- 614. doi: 10.1016/j.ins.2021.07.048
30	孟勤超, 杨翠丽, 乔俊飞. 基于改进SPEA2算法的给水管网多目标优化设计[J]. 智能系统学报, 2018, 13 (1): 118- 124.
	MENG Q C , YANG C L , QIAO J F . Multi-objective optimization design of water distribution systems based on improved SPEA2 algorithm[J]. CAAI Transactions on Intelligent Systems, 2018, 13 (1): 118- 124.
31	GU Q H , BAI J M , LI X X , et al. A constrained multi-objective evolutionary algorithm based on decomposition with improved constrained dominance principle[J]. Swarm and Evolutionary Computation, 2022, 75, 101162. doi: 10.1016/j.swevo.2022.101162
32	华东师范大学. 一种基于罚函数交叉边界法的多目标自适应分类方法及系统[P]. 中国: CN202110884575.4, 2021-11-23.
	East China Normal University. A multi object adaptive classification method and system based on penalty function cross boundary method[P]. China: CN202110884575.4, 2021-11-23.
33	SEBASTIÁN L , LORENA P , ⅡRO P , et al. Multi objective optimization of a continuous kraft pulp digester using SPEA2[J]. Computers & Chemical Engineering, 2020, 143, 107086.
34	孙克乙. 基于多任务迁移演化学习的超多目标优化[D]. 大连: 大连理工大学, 2019.
	SUN K Y. Many-objective optimization based on evolutionary multi-tasking and transfer learning[D]. Dalian: Dalian University of Technology, 2019.

参数符号	符号描述
n	数据类型数量
x_j	采集成员j的任务负载量(1≤j≤m)
m	负责采集的联邦成员数量
L(K_i)	第i类数据对象的数据量(1≤i≤n) 所有采集成员的平均任务负载量
${\bar L}$	$\bar{L}=\frac{\sum\limits_{i=1}^n L\left(K_i\right)}{m}$
v	采集成员的处理速度
t	一个仿真步长
T	仿真步长数
D	采集成员集合D={N_j\|1≤j≤m}
T	数据对象类型集合T={K_i\|1≤i≤n}

参数	取值
交叉概率	1
模拟二进制交叉分布指数	20
多项式变异的概率	1/200
变异非分布指数	20

参数	取值
惯性权重	0.4
压缩因子	2.05
加速系数	1至数据类型数之间的随机数

参数	取值
交叉概率	1/200
差分进化交叉分布指数	20
多项式变异的概率	1
变异非分布指数	20

算例	HV值				算法耗时/s
算例	MOEA-D	MOPSO	SPEA-2	NSGA-Ⅱ	MOEA-D	MOPSO	SPEA-2	NSGA-Ⅱ
1	0.058(0.161 2)-	0.55(2.22e-16)≈	0.510(0.134 0)≈	0.49(0.161 2)	65.59	2.14	10.56	0.93
2	0.160(0.111 0)-	0.60(0.135 0)≈	0.526(0.072 6)≈	0.52(0.072 2)	92.54	19.68	25.04	1.07
3	0.190(0.037 0)-	0.44(0.040 7)+	0.300(0.031 2)+	0.29(0.024 2)	76.04	4.31	1.95	1.24