基于宏观-微观层次化仿真模型的动态数据驱动仿真方法

doi:10.16182/j.issn1004731x.joss.24-0787

摘要/Abstract

摘要：

本文提出一种基于宏观-微观层次化仿真模型的动态数据驱动仿真方法。这种仿真方法可以使来自真实系统的观测数据逐次影响到宏观仿真和微观仿真，实现宏观仿真和微观仿真的协同演进，从而既可以为决策者提供真实系统宏观层面的演化趋势预测，为辅助决策提供依据，又可以为决策者提供与真实系统相似的微观仿真试验台，支撑策略的仿真推演和量化评估。建立了该方法的形式化描述，设计了适用于宏观-微观层次化仿真模型的数据同化方法；设计了一个疫情爆发和传播的仿真实验场景，采用identical-twin仿真实验方法，验证了本文所提方法的可行性和有效性。

关键词: 动态数据驱动仿真, 宏观-微观层次化仿真, 数据同化, 传染病传播, 仿真与预测

Abstract:

This paper proposed a dynamic data driven simulation approach based on macro-microscopic hierarchical simulation models. This approach enabled the measurement data from the real system to affect the macroscopic simulation and microscopic simulation in sequence and made the two simulations evolve together so that it could provide decision makers with the state evolution prediction of the real system at the macroscopic level to assist decision making and provide a microscopic testbed similar to the real system, on which decision makers could deduce and evaluate their strategies. This paper established a formal description of the approach and designed a data assimilation method applicable to macro-microscopic hierarchical simulation models. A simulated epidemic outbreak and transmission scenario was designed, based on which an identical-twin experiment was conducted to illustrate the feasibility and effectiveness of the proposed approach.

Key words: dynamic data driven simulation, macro-microscopic hierarchical simulation, data assimilation, epidemic transmission, simulation and prediction

中图分类号:

TP391.9

谢旭,马雨晴 . 基于宏观-微观层次化仿真模型的动态数据驱动仿真方法[J]. 系统仿真学报, 2025, 37(7): 1848-1864.

Xie Xu,Ma Yuqing . Dynamic Data Driven Simulation Based on Macro-microscopic Hierarchical Simulation Models[J]. Journal of System Simulation, 2025, 37(7): 1848-1864.

图/表 16

图1

表1

图2

图3

图4

表2

图5

图6

表3

图7

表4

表5

表6

模型误差对预测性能的影响

模型误差		$E D A, S$	$E D A, E$	$E D A, I$	$E D A, R$
5%	macro	89.76 (88.67, 90.02)	49.07 (48.68, 49.26)	22.25 (22.09, 22.33)	66.17 (64.53, 66.54)
5%	micro	7.29 (6.84, 7.39)	3.62 (3.52, 3.79)	2.37 (2.31, 2.49)	6.69 (6.61, 7.10)
10%	macro	95.56 (94.90, 95.63)	50.36 (50.27, 50.53)	22.89 (22.85, 22.94)	71.42 (70.83, 71.46)
10%	micro	11.40 (11.16, 11.57)	5.07 (4.93, 5.23)	3.57 (3.45, 3.68)	12.64 (12.20, 12.80)
15%	macro	101.51 (100.05, 101.98)	51.67 (51.42, 51.74)	23.42 (23.33, 23.50)	76.47 (75.56, 77.00)
15%	micro	15.71 (14.43, 16.98)	6.10 (5.85, 6.36)	4.49 (4.28, 4.72)	16.48 (16.26, 16.71)

表6

表7

数据误差对预测性能的影响

数据误差 $σ$		$E D A, S$	$E D A, E$	$E D A, I$	$E D A, R$
3	macro	94.96 (94.59, 95.21)	49.68 (48.53, 50.43)	22.40 (21.25, 22.88)	71.00 (70.46, 71.28)
3	micro	10.35 (10.06, 10.62)	4.72 (4.61, 4.77)	3.33 (3.22, 3.40)	11.38 (11.02, 11.62)
5	macro	95.56 (94.90, 95.63)	50.36 (50.27, 50.53)	22.89 (22.85, 22.94)	71.42 (70.83, 71.46)
5	micro	11.40 (11.16, 11.57)	5.07 (4.93, 5.23)	3.57 (3.45, 3.68)	12.64 (12.20, 12.80)
7	macro	96.87 (95.35, 97.11)	50.82 (50.52, 51.02)	23.09 (22.94, 23.14)	72.42 (71.74, 72.97)
7	micro	11.57 (11.33, 11.92)	5.28 (5.19, 5.35)	3.65 (3.55, 3.73)	12.71 (12.49, 13.11)

表7

表8

数据采集频率对预测性能的影响

数据采集频率 $Δ T$		$E D A, S$	$E D A, E$	$E D A, I$	$E D A, R$
3	macro	92.60 (91.79, 93.52)	48.08 (47.80, 48.29)	21.57 (21.40, 21.76)	69.30 (68.49, 70.00)
3	micro	11.09 (10.66, 11.17)	5.03 (4.86, 5.28)	3.53 (3.28, 3.71)	12.08 (11.74, 12.55)
4	macro	95.56 (94.90, 95.63)	50.36 (50.27, 50.53)	22.89 (22.85, 22.94)	71.4183 (70.83, 71.46)
4	micro	11.40 (11.16, 11.57)	5.07 (4.93, 5.23)	3.57 (3.45, 3.68)	12.64 (12.20, 12.80)
5	macro	99.97 (98.73, 100.72)	51.18 (50.98, 51.30)	23.29 (23.17, 23.34)	74.48 (73.14, 74.76)
5	micro	11.56 (11.27, 11.84)	5.20 (4.94, 5.71)	3.57 (3.49, 3.65)	12.70 (12.41, 13.25)

表8

表9

微观仿真数量对预测性能的影响

微观仿真数量 $N 2$	$E m i c r o, D A, S$	$E m i c r o, D A, E$	$E m i c r o, D A, I$	$E m i c r o, D A, R$
100	11.60 (10.61, 11.98)	5.23 (4.98, 5.42)	3.69 (3.50, 3.87)	12.92 (12.07, 13.40)
300	11.40 (11.16, 11.57)	5.07 (4.93, 5.23)	3.57 (3.45, 3.68)	12.64 (12.20, 12.80)
500	11.06 (10.90, 11.27)	4.88 (4.72, 4.97)	3.44 (3.26, 3.51)	12.03 (11.79, 12.32)

表9

参考文献 35

[1]	Hu X, Long Y. Dynamic Data Driven Simulation[J]. SCS M & S Magazine, 2011, II(1): 16-22.
[2]	Bouttier F, Courtier P. Data Assimilation Concepts and Methods[EB/OL]. [2024-05-18]. .
[3]	Xue Haidong, Gu Feng, Hu Xiaolin. Data Assimilation Using Sequential Monte Carlo Methods in Wildfire Spread Simulation[J]. ACM Transactions on Modeling and Computer Simulation, 2012, 22(4): 23.
[4]	Wang Minghao, Hu Xiaolin. Data Assimilation in Agent Based Simulation of Smart Environments Using Particle Filters[J]. Simulation Modelling Practice and Theory, 2015, 56: 36-54.
[5]	Xie Xu, van Lint Hans, Verbraeck Alexander. A Generic Data Assimilation Framework for Vehicle Trajectory Reconstruction on Signalized Urban Arterials Using Particle Filters[J]. Transportation Research Part C: Emerging Technologies, 2018, 92: 364-391.
[6]	Rai S, Hu Xiaolin. Building Occupancy Simulation and Data Assimilation Using a Graph-based Agent-oriented Model[J]. Physica A: Statistical Mechanics and Its Applications, 2018, 502: 270-287.
[7]	Hu Xiaolin. Dynamic Data-Driven Simulation: Real-time Data for Dynamic System Analysis and Prediction[M]. Singapore: World Scientific Publishing Co Pte Ltd, 2023.
[8]	Xie Xu. Data Assimilation in Discrete Event Simulations[D]. Delft: Delft University of Technology, 2018.
[9]	谢旭, 邱晓刚, 包亦正, 等. 动态数据驱动仿真综述[J]. 系统仿真学报, 2024, 36(8): 1832-1842.
	Xie Xu, Qiu Xiaogang, Bao Yizheng, et al. Dynamic Data Driven Simulation: An Overview[J]. Journal of System Simulation, 2024, 36(8): 1832-1842.
[10]	Lahoz William A, Schneider Philipp. Data Assimilation : Making Sense of Earth Observation[J]. Frontiers in Environmental Science, 2014, 2: 00016.
[11]	Wu Wanshu, James Purser R, Parrish David F. Three-dimensional Variational Analysis with Spatially Inhomogeneous Covariances[J]. Monthly Weather Review, 2022, 130(12): 2905-2916.
[12]	Lorenc A C, Rawlins F. Why Does 4D-var beat 3D-var?[J]. Quarterly Journal of the Royal Meteorological Society, 2005, 131(613): 3247-3257.
[13]	Arulampalam M S, Maskell S, Gordon N, et al. A Tutorial on Particle Filters for Online Nonlinear/Non-gaussian Bayesian Tracking[J]. IEEE Transactions on Signal Processing, 2002, 50(2): 174-188.
[14]	Gillijns S, Mendoza O B, Chandrasekar J, et al. What Is the Ensemble Kalman Filter and How Well Does It Work?[C]//Proceedings of the 2006 American Control Conference. Piscataway: IEEE, 2006: 4448-4453.
[15]	Evensen Geir. The Ensemble Kalman Filter: Theoretical Formulation and Practical Implementation[J]. Ocean Dynamics, 2003, 53(4): 343-367.
[16]	Djuric P M, Kotecha J H, Zhang J, et al. Particle Filtering[J]. IEEE Signal Processing Magazine, 2003, 20(5): 19-38.
[17]	Peter Jan van Leeuwen. Particle Filtering in Geophysical Systems[J]. Monthly Weather Review, 2009, 137(12): 4089-4114.
[18]	Aydt Heiko, Stephen John Turner, Cai Wentong, et al. Symbiotic Simulation Systems: An Extended Definition Motivated by Symbiosis in Biology[C]//2008 22nd Workshop on Principles of Advanced and Distributed Simulation. Piscataway: IEEE, 2008: 109-116.
[19]	Darema F. Dynamic Data Driven Applications Systems: A New Paradigm for Application Simulations and Measurements[C]//Computational Science - ICCS 2004. Berlin: Springer Berlin Heidelberg, 2004: 662-669.
[20]	Darema F. Dynamic Data Driven Applications Systems: New Capabilities for Application Simulations and Measurements[C]//Computational Science – ICCS 2005. Berlin: Springer Berlin Heidelberg, 2005: 610-615.
[21]	Kamrani Farzad, Ayani Rassul. Using On-line Simulation for Adaptive Path Planning of UAVs[C]//11th IEEE International Symposium on Distributed Simulation and Real-Time Applications (DS-RT'07). Piscataway: IEEE, 2007: 167-174.
[22]	段伟. 平行仿真的内涵、发展与应用[J]. 指挥与控制学报, 2019, 5(2): 82-86.
	Duan Wei. Parallel Simulation: Motivation, Concept and Application[J]. Journal of Command and Control, 2019, 5(2): 82-86.
[23]	王会霞. 平行仿真技术研究[J]. 航天控制, 2016, 34(6): 64-67.
	Wang Huixia. Research on Parallel Simulation Technology[J]. Aerospace Control, 2016, 34(6): 64-67.
[24]	窦林涛, 初阳, 周玉芳, 等. 平行仿真技术在指控系统中的应用构想[J]. 指挥控制与仿真, 2017, 39(1): 62-69.
	Dou Lintao, Chu Yang, Zhou Yufang, et al. Conception of the Application of Parallel Simulation Technology in Command and Control System[J]. Command Control & Simulation, 2017, 39(1): 62-69.
[25]	Grieves M, Vickers J. Digital Twin: Mitigating Unpredictable, Undesirable Emergent Behavior in Complex Systems[M]//Franz-Josef Kahlen, Flumerfelt S, Anabela Alves. Transdisciplinary Perspectives on Complex Systems: New Findings and Approaches. Cham: Springer International Publishing, 2017: 85-113.
[26]	张霖. 关于数字孪生的冷思考及其背后的建模和仿真技术[J]. 系统仿真学报, 2020, 32(4): 1-10.
[27]	闫雪飞, 李新明, 刘东, 等. 基于多分辨率的multi-Agent武器装备体系作战仿真研究[J]. 系统仿真学报, 2017, 29(1): 136-143, 153.
	Yan Xuefei, Li Xinming, Liu Dong, et al. Battle Simulation Study on Multi-agent Weapon System-of-systems Based on Multi-resolution[J]. Journal of System Simulation, 2017, 29(1): 136-143, 153.
[28]	刘宝宏, 黄柯棣. 多分辨率建模的研究现状与发展[J]. 系统仿真学报, 2004, 16(6): 1150-1154.
	Liu Baohong, Huang Kedi. Multi-resolution Modeling: Present Status and Trends[J]. Journal of System Simulation, 2004, 16(6): 1150-1154.
[29]	刘兆鹏, 徐新海, 袁博文, 等. 一种多分辨率仿真模型构建方法[J]. 系统仿真学报, 2023, 35(1): 202-211.
	Liu Zhaopeng, Xu Xinhai, Yuan Bowen, et al. A Multi-resolution Simulation Modeling Method[J]. Journal of System Simulation, 2023, 35(1): 202-211.
[30]	毛华超, 彭功状, 张和明, 等. 面向模型粒度控制的多分辨率建模方法[J]. 系统仿真学报, 2015, 27(11): 2630-2637.
	Mao Huachao, Peng Gongzhuang, Zhang Heming, et al. Model Granularity Control Oriented Multi-resolution Modeling[J]. Journal of System Simulation, 2015, 27(11): 2630-2637.
[31]	张高峰, 吉玉洁. 多分辨率模型的协同仿真方法研究[J]. 系统仿真学报, 2018, 30(7): 2525-2532.
	Zhang Gaofeng, Ji Yujie. Method Research on Collaborative Simulation of Multi-resolution Models[J]. Journal of System Simulation, 2018, 30(7): 2525-2532.
[32]	Treiber Martin, Kesting Arne. Traffic Flow Dynamics: Data, Models and Simulation[M]. Berlin: Springer Berlin Heidelberg, 2013.
[33]	陈海亮. 基于地理信息的大规模疾病传播仿真与防控措施评估研究[D]. 长沙: 国防科技大学, 2021.
	Chen Hailiang. Research on Simulation of Large-scale Disease Spread and Evaluation of Mitigation Strategies Based on Geographic Information[D]. Changsha: National University of Defense Technology, 2021.
[34]	Ramesh Chandra Poonia, Abdul Khader Jilani Saudagar, Altameem Abdullah, et al. An Enhanced SEIR Model for Prediction of COVID-19 with Vaccination Effect[J]. Life, 2022, 12(5): 647.
[35]	Yang Xian, Wang Shuo, Xing Yuting, et al. Bayesian Data Assimilation for Estimating Instantaneous Reproduction Numbers During Epidemics: Applications to COVID-19[J]. PLOS Computational Biology, 2022, 18(2): e1009807.

街道	总人口	中心经度/(°)	中心纬度/(°)
民族街道	32 213	114.285	30.570 5
花楼街道	2 354	114.29	30.578 3
水塔街道	19 426	114.282	30.583 7
民权街道	36 387	114.287	30.575 3
满春街道	23 878	114.278	30.573
民意街道	35 693	114.274	30.579 8
新华街道	40 765	114.272	30.589
万松街道	83 040	114.246	30.596
唐家墩街道	107 612	114.268	30.614 8
北湖街道	37 768	114.265	30.602 3
前进街道	29 010	114.279	30.581 4
常青街道	86 882	114.249	30.613 4
汉兴街道	150 671	114.242	30.627 2

感染状态	no-DA	DA	误差下降/%
S	234.73	95.24	59.43
E	99.72	50.38	49.48
I	69.90	22.89	67.25
R	275.24	71.40	74.06

感染状态	no-DA	DA	误差下降/%
S	18.92	11.37	39.90
E	8.57	5.14	40.02
I	5.96	3.58	39.93
R	22.32	12.55	43.77

仿真方法	样本数	预测准确度RMSE				运行时间/s
仿真方法	样本数	S	E	I	R	运行时间/s
多样本仿真	100	18.92	8.57	5.96	22.32	/
传统DDDS	100	8.56	3.82	2.73	9.99	798.44
	300	8.18	3.71	2.62	9.44	2 031.23
	500	7.48	3.35	2.38	8.75	2 890.48
基于层次化仿真模型的DDDS	100	11.73	5.38	3.75	12.90	784.44
	300	11.37	5.14	3.58	12.55	2 035.82
	500	11.04	4.67	3.18	11.49	2 783.57