融合多智能体与超图的复杂动态系统建模方法探索

doi:10.11996/JG.j.2095-302X.2023030599

图学学报 ›› 2023, Vol. 44 ›› Issue (3): 599-608.DOI: 10.11996/JG.j.2095-302X.2023030599

• 计算机图形学与虚拟现实 • 上一篇下一篇

融合多智能体与超图的复杂动态系统建模方法探索

王鹏飞¹(), 陶体伟², 焦点¹, 申彦明¹, 周东生³, 张强¹()

1.大连理工大学计算机科学与技术学院，辽宁大连 116024
2.泰山学院信息科学技术学院，山东泰安 271000
3.大连大学软件学院，辽宁大连 116622

收稿日期:2022-08-26 接受日期:2022-12-16 出版日期:2023-06-30 发布日期:2023-06-30
通讯作者: 张强(1971-)，男，教授，博士。主要研究方向为复杂系统。E-mail：zhangq@dlut.edu.cn
作者简介:
王鹏飞(1990-)，男，副教授，博士。主要研究方向为复杂系统、多模态智能计算。E-mail：wangpf@dlut.edu.cn
基金资助:
科技创新2030—“新一代人工智能”重大项目(2021ZD0112400);国家自然科学基金委-辽宁联合基金项目(U1908214);中央高校基本科研基金项目(DUT21TD107);中央高校基本科研基金项目(DUT20RC(3)039);辽宁省兴辽人才计划项目(XLYC2008017);CCF-腾讯犀牛鸟基金项目(IAGR20210116)

Exploration on the modeling method of complex dynamic system integrating multi-agent and hypergraph

WANG Peng-fei¹(), TAO Ti-wei², JIAO Dian¹, SHEN Yan-ming¹, ZHOU Dong-sheng³, ZHANG Qiang¹()

1. School of Computer Science and Technology, Dalian University of Technology, Dalian Liaoning 116024, China
2. School of Information Science and Technology, TaiShan University, Tai’an Shandong 271000, China
3. School of Software, Dalian University, Dalian Liaoning 116622, China

Received:2022-08-26 Accepted:2022-12-16 Online:2023-06-30 Published:2023-06-30
Contact: ZHANG Qiang (1971-), professor, Ph.D. His main research interest covers complex systems. E-mail：zhangq@dlut.edu.cn
About author:
WANG Peng-fei (1990-), associate professor, Ph.D. His main research interests cover complex systems, multimodal intelligent computing. E-mail：wangpf@dlut.edu.cn
Supported by:
National Key Research and Development Program of China(2021ZD0112400);NSFC-Liaoning Province United Foundation(U1908214);Fundamental Research Funds for the Central Universities(DUT21TD107);Fundamental Research Funds for the Central Universities(DUT20RC(3)039);Liaoning Revitalization Talents Program(XLYC2008017);CCF-Tencent Open Fund(IAGR20210116)

摘要/Abstract

摘要：

在自然界以及人类社会中，绝大多数系统本质上都可以抽象表示为复杂系统，针对当前复杂系统复杂性不断增加这一难题，亟需完善且成熟的复杂系统理论与方法进行建模研究与处理。当前基于图的复杂系统建模方法难以描绘节点间极其复杂的连接以及节点间的高阶关系，同时也难以对复杂系统的智能化感知、决策和控制等行为进行有效刻画。基于此，提出一种融合多智能体与超图的复杂动态系统建模方法，模型从几种不同的演化角度进行动态演化，对复杂动态系统进行了具体的描绘。此模型通过赋予复杂系统中个体智能化特征，进行感知、决策和控制，同时也能够更好地描绘智能体节点间的高阶关系，为复杂系统的智能理论研究提供新思路和新方法。

关键词: 多智能体, 超图, 复杂系统, 建模, 复杂网络, 超网络

Abstract:

Most systems can be abstracted as complex systems in nature and human society. Given the increasing complexity of today’s complex systems, there is an urgent need for advanced and mature complex system theories and methods for modeling research and processing. However, the current graph-based modeling methods used in complex systems encounter difficulties in depicting the extremely complex connections between nodes and the higher-order relationships between them. Additionally, these methods face challenges in effectively portraying the intelligent perception, decision-making, and control of complex systems. A modeling method for complex dynamic systems integrating multi-agents and hypergraphs was proposed to address these issues. This model dynamically evolved from several different evolutionary angles to specifically describe complex dynamic systems. This model enabled perception, decision-making, and control by conferring agent nodes with intelligent features in complex systems and better describing the higher-order relationships between agent nodes. As a result, this modeling approach provided novel ideas and methods for the study of the intelligence theory of complex systems.

Key words: multi-agents, hypergraphs, complex systems, modeling, complex networks, hypernetworks

中图分类号:

TP391

王鹏飞, 陶体伟, 焦点, 申彦明, 周东生, 张强. 融合多智能体与超图的复杂动态系统建模方法探索[J]. 图学学报, 2023, 44(3): 599-608.

WANG Peng-fei, TAO Ti-wei, JIAO Dian, SHEN Yan-ming, ZHOU Dong-sheng, ZHANG Qiang. Exploration on the modeling method of complex dynamic system integrating multi-agent and hypergraph[J]. Journal of Graphics, 2023, 44(3): 599-608.

图/表 19

图1 多智能体系统

Fig. 1 Multi-agent system

图2 传统图和超图((a)传统图；(b)超图)

Fig. 2 Traditional graph and hypergraph ((a) Traditional graph; (b) Hypergraph)

图3 融合多智能体与超图的模型

Fig. 3 A model that integrating multi-agent and hypergraph

图4 融合多智能体与超图的交通模型

Fig. 4 A traffic model that integrating multi-agent and hypergraph

图5 模型初始状态

Fig. 5 The initial state of the model

图6 模型演化过程(1)

Fig. 6 The evolution process (1) of the model

图7 模型演化过程(2)

Fig. 7 The evolution process (2) of the model

图8 模型演化过程(3)((a)添加新节点；(b)选取孤立节点)

Fig. 8 The evolution process (3) of the model ((a) Add new node; (b) Pick orphaned node)

图9 模型演化过程(4)

Fig. 9 The evolution process (4) of the model

图10 上海出租车路径轨迹((a)附带卫星地图；(b)不附带卫星地图)

Fig. 10 Taxi paths in Shanghai ((a) With satellite map; (b) Without satellite map)

图11 简化超图交通模型

Fig. 11 Simplified hypergraph traffic model

图12 标志节点的超图交通模型(局部)

Fig. 12 A hypergraph traffic model with marked nodes (local)

图13 节点超度分布

Fig. 13 The distribution of node hyperdegree

图14 超边度分布

Fig. 14 The distribution of hyperedge degree

图15 节点强度分布

Fig. 15 The distribution of node intensity

图16 节点超度与节点强度的关系

Fig. 16 The relationship of node hyperdegree and node intensity

图17 超边强度分布

Fig. 17 The distribution of hyperedge intensity

图18 超边度与超边强度的关系

Fig. 18 The relationship between hyperedge degree and hyperedge intensity

图19 超边基数分布

Fig. 19 The distribution of hyperedge cardinality

参考文献 22

[1]	李杨, 徐峰, 谢光强, 等. 多智能体技术发展及其应用综述[J]. 计算机工程与应用, 2018, 54(9): 13-21. DOI
	LI Y, XU F, XIE G Q, et al. Survey of development and application of multi-agent technology[J]. Computer Engineering and Applications, 2018, 54(9): 13-21. (in Chinese) DOI
[2]	孙智诚. 考虑驾驶行为演化的多智能体交通流模型研究[D]. 西安: 长安大学, 2020.
	SUN Z C. Research on multi-agent traffic flow model considering driving behavior evolution[D]. Xi’an: Chang’an University, 2020. (in Chinese)
[3]	陈磊, 李钟慎. 多智能体系统一致性综述[J]. 自动化博览, 2018, 35(2): 74-78.
	CHEN L, LI Z S. Literature review on the consistency of multi-agent systems[J]. Automation Panorama, 2018, 35(2): 74-78. (in Chinese)
[4]	YAMADA M, INOKUCHI A. Similar supergraph search based on graph edit distance[J]. Algorithms, 2021, 14(8): 225. DOI URL
[5]	GHARAHIGHEHI A. Fair multi-stakeholder news recommender system with hypergraph ranking[J]. Information Processing & Management, 2021, 58(5): 102663. DOI URL
[6]	PENG J, ZHANG B, SUGENG K A. Uncertain hypergraphs: a conceptual framework and some topological characteristics indexes[J]. Symmetry, 2022, 14(2): 330. DOI URL
[7]	GAO Y, ZHANG Z Z, LIN H J, et al. Hypergraph learning: methods and practices[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(5): 2548-2566.
[8]	SHAO S, WANG M K, XU R, et al. DLDL: dynamic label dictionary learning via hypergraph regularization[J]. Neurocomputing, 2022, 475: 80-88. DOI URL
[9]	HEWASINGHAGE M, VARGA J, ABELLÓ A, et al. Managing polyglot systems metadata with hypergraphs[M]// Conceptual Modeling. Cham: Springer International Publishing, 2018: 463-478.
[10]	罗海秀, 赵海兴, 肖玉芝, 等. 基于超图的公交超网络拓扑特性及鲁棒性分析[J]. 西南大学学报: 自然科学版, 2021, 43(10): 181-191.
	LUO H X, ZHAO H X, XIAO Y Z, et al. A hypergraph-based analysis of the topology and robustness of bus hypernetworks[J]. Journal of Southwest University: Natural Science Edition, 2021, 43(10): 181-191. (in Chinese)
[11]	YIN N, FENG F L, LUO Z G, et al. Dynamic hypergraph convolutional network[C]// 2022 IEEE 38th International Conference on Data Engineering. New York: IEEE Press, 2022: 1621-1634.
[12]	YU P, WANG Z P, WANG P W, et al. Dynamic evolution of shipping network based on hypergraph[J]. Physica A: Statistical Mechanics and its Applications, 2022, 598: 127247. DOI URL
[13]	LI M H, PETERS C, LEIMEISTER J. Moving towards a non-dyadic view on service systems and its operationalization - applying the hypergraph-based service system model[C]// The 53rd Hawaii International Conference on System Sciences. Honolulu: University of Hawaii, 2020: 1600-1607.
[14]	WANG Z X. A hypergraph-based approach for context-aware smart product-service system configuration[J]. Computers & Industrial Engineering, 2022, 163: 107816. DOI URL
[15]	HAMIDI M, SAEID A B. Secreted graphs to binary codes and applications[J]. Soft Computing, 2022, 26(11): 5025-5042. DOI
[16]	ZHEN Y M, WANG J H. Community detection in general hypergraph via graph embedding[J]. Journal of the American Statistical Association, 2021: 1-10.
[17]	DENG Q W, ZHANG S Y, DING Z. An efficient hypergraph approach to robust point cloud resampling[J]. IEEE Transactions on Image Processing: a Publication of the IEEE Signal Processing Society, 2022, 31: 1924-1937. DOI URL
[18]	索琪. 超网络的结构、演化及传播动力学研究[D]. 上海: 上海理工大学, 2018.
	SUO Q. The structure, evolving models, and spreading dynamics of hypernetworks[D]. Shanghai: University of Shanghai for Science & Technology, 2018. (in Chinese)
[19]	LUO Y W, ZHENG L, GUAN T, et al. Taking a closer look at domain shift: category-level adversaries for semantics consistent domain adaptation[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition. New York: IEEE Press, 2019: 2502-2511.
[20]	LUO Y W, LIU P, ZHENG L, et al. Category-level adversarial adaptation for semantic segmentation using purified features[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(8): 3940-3956.
[21]	LUO Y W, LIU P, GUAN T, et al. Adversarial style mining for one-shot unsupervised domain adaptation[C]// The 34th International Conference on Neural Information Processing Systems. New York: ACM, 2020: 20612-20623.
[22]	LU Y L, LUO Y W, ZHANG L, et al. Bidirectional self-training with multiple anisotropic prototypes for domain adaptive semantic segmentation[EB/OL]. [2022-01-02]. https://arxiv.org/abs/2204.07730.

融合多智能体与超图的复杂动态系统建模方法探索

Exploration on the modeling method of complex dynamic system integrating multi-agent and hypergraph

RichHTML

PDF (PC)

可视化

摘要/Abstract

引用本文

使用本文

图/表 19

参考文献 22

相关文章 15

编辑推荐

Metrics

本文评价

[1]	邹强. 浅谈实体建模：历史、现状与未来[J]. 图学学报, 2022, 43(6): 987-1001.
[2]	陈天翔, 陈斌. 一种基于全景视频的 6 自由度野外虚拟环境快速构建方法[J]. 图学学报, 2022, 43(5): 901-908.
[3]	袁泽, 陈斌. 基于 INLA-SPDE 方法的区域污染物模拟与预测[J]. 图学学报, 2022, 43(1): 125-132.
[4]	薛靖国, 侯学良. 基于 BIM+CV 的现浇混凝土建筑结构构件定位 [J]. 图学学报, 2022, 43(1): 156-162.
[5]	张文元, 谈国新. 建筑物多尺度三维语义建模研究[J]. 图学学报, 2022, 43(1): 163-171.
[6]	杨延璞, 兰晨昕, 雷紫荆, 王欣蕊, 龚政. 产品造型设计多阶段网络耦合决策方法研究 [J]. 图学学报, 2021, 42(6): 1018-1026.
[7]	张玥焜, 余文杰, 赵习之, 吕艳星, 冯伟桓, 李峥嵘, 胡少军 . 基于机载激光雷达点云的交互式树木分割与建模方法研究[J]. 图学学报, 2021, 42(4): 599-607.
[8]	张欣悦 , 雷一凡 , 刘培培 , 包芳勋 , 张云峰 . 可变参数的有理分形插值曲线建模[J]. 图学学报, 2021, 42(2): 245-255.
[9]	肖文磊, 王志明, 王世平, 赵罡. 基于周期性晶格的 3D 打印模型轻量化方法[J]. 图学学报, 2021, 42(2): 263-270.
[10]	张雪才 , 陈丽晔 , 王正中 . 水工弧形闸门结构的 APDL 建模方法[J]. 图学学报, 2021, 42(2): 271-278.
[11]	冯洁 , 李博 , 周秉锋 , . 基于像素聚类的空间变化表面材质建模[J]. 图学学报, 2021, 42(1): 94-100.
[12]	张航, 张树生, 杨磊. 基于深度学习的孔特征可制造性分析方法 [J]. 图学学报, 2021, 42(1): 117-123.
[13]	王宇昆1,2，曹力 1,2，赵洋 1,2，李琳 1,2. 基于有向无环图的立交结构构建方法[J]. 图学学报, 2020, 41(4): 583-592.
[14]	张肇轩 1，王诚斌 1，杨鑫 1，朴星霖 2，王鹏杰 3，尹宝才 1 . 基于模板替换的室内场景建模方法研究[J]. 图学学报, 2020, 41(2): 270-276.
[15]	林莹莹，蔡睿凡，朱雨真，唐祥峻，金小刚. 基于Leap Motion 的虚拟现实陶艺体验系统[J]. 图学学报, 2020, 41(1): 57-65.