To select the most credible simulation model among multiple alternatives with dynamic and correlated outputs, an intelligent model selection method based on graph representation learning (GRL) is proposed, which treats the model selection problem as a reference data categorization problem. To model the correlations between the output variables, the evaluation data are initially transformed from their traditional two-dimensional format into a graph-structured format based on the distance correlation coefficients. Then, graph isomorphism networks (GINs) are employed to achieve GRL and evaluation data classification. The graph embeddings produced by GRL, which represent the interdependencies and dynamic evolutionary patterns among variables, enable the categorization of reference data as originating from one of the alternative models. Finally, the most credible simulation model is determined based on the category probabilities of the multisample reference data. The effectiveness of the proposed method in feature extraction and model selection is demonstrated through an application example on aerodynamic parameter models of a flight vehicle.
RobinsonS.Simulation: the practice of model development and use. 2nd ed.Basingstoke: Palgrave Macmillan, 2014, pp. 1–5.
2.
LawAWKeltonWD.Simulation modeling and analysis. 3rd ed.New York: McGraw Hill, 2000, pp. 1–3.
3.
LiWLinSLZhouYC, et al. Research progress on credibility assessment of a complex simulation system. Sci Sin Inform2018; 48: 767–782.
4.
SargentRG.Verification and validation of simulation models. J Simul2013; 7: 12–24.
5.
DurstPJAndersonDTBethelCL.A historical review of the development of verification and validation theories for simulation models. Int J Model Simul Sci Comput2017; 8: 1730001.
6.
GoldsmanD.Ranking and selection in simulation. In: Proceedings of the winter simulation conference (WSC), Virginia, 12–14 December 1983, pp.387–393. New York: IEEE.
7.
HaoGLiuLLiG.Polynomial regression-based metamodeling technique for global approximation in simulation-based engineering design. Eng Computation2013; 30: 762–779.
8.
ZhangCHanS.Nonlinear regression analysis applied to dynamic simulation modeling of complex systems. Simul Model Pract Th2012; 26: 12–24.
9.
GaurSGuptaR.Neural network-based modeling and simulation of a nonlinear system for control. J Process Contr2012; 22: 47–56.
10.
LiWJiaoSLuLY, et al. Validation and selection of simulation model based on the feature differences. Acta Automat Sin2014; 40: 2134–2144.
11.
Van AckerBDe MeulenaerePVangheluweH, et al. Validity frame–enabled model-based engineering processes. Simulation2024; 100: 185–226.
12.
LawAM.How to build valid and credible simulation models. In: Proceedings of the winter simulation conference (WSC), Singapore, 11–14 December 2022, pp. 1283–1295. New York: IEEE.
13.
ZhouYC.Research on validation method for complex simulation model. PhD Thesis, Harbin Institute of Technology, Harbin, China, 2019.
14.
BensonKCGoldsmanDPritchettAR. Ranking and selection procedures for simulation. In: Proceedings of the winter simulation conference (WSC), Monterey, CA, 3–6 December 2006, pp. 179–185 New York: IEEE.
15.
HongLJFanWLuoJ.Review on ranking and selection: a new perspective. Front Eng Manag2021; 8: 321–343.
16.
ZhengKHuJZhanZF, et al. Multivariate responses analysis for model validation of dynamic systems. J Shanghai Jiaotong Univ2015; 49: 191–195.
17.
DongHHeRJinM. Power system dynamic simulation validation based on similarity theory and analytical hierarchy process. In: Proceedings of the 2006 international conference on power system technology, Chongqing, China, 22–26 October 2006, pp. 1–7. New York: IEEE.
18.
XiZMHaoPFuY, et al. Validation metric for dynamic system responses under uncertainty. SAE Int J Mater Manuf2015; 8: 309–314.
19.
ZhanZFFuYYangR, et al. Bayesian based multivariate model validation method under uncertainty for dynamic systems. J Mech Des: T ASME2012; 134: 034502.
20.
MartensJPauwelsKPutF. A neural network approach to the validation of simulation models. In: Proceedings of the 2006 winter simulation conference (WSC), Monterey, CA, USA, 3–6 December 2006, pp. 250–256. New York: IEEE.
21.
ZhengZZGuoCCChenJY, et al. Graph Neural Network-based representation learning for medical time series. In: Proceedings of the 32nd international conference on artificial neural networks (ICANN), Heraklion, 26–29 September 2023, vol. 14259, pp. 194–205. Cham: Springer.
22.
GengXLHeXYXuLY.Graph correlated attention recurrent neural network for multivariate time series forecasting. Inform Sciences2022; 606: 126–142.
23.
ZhouLWZengQKLiB.Hybrid anomaly detection via multihead dynamic graph attention networks for multivariate time series. IEEE Access2022; 10: 40967–40978.
24.
RebbaRMahadevanS.Validation of models with multivariate output. Reliab Eng Syst Safe2006; 91: 861–871.
25.
LinSLLiWYangM, et al. Multivariate validation method under correlation for simulation model. Acta Automat Sin2019; 45: 1666–1678.
26.
WangHY.Research on multivariate simulation result validation methods under uncertainty. PhD Thesis, Harbin Institute of Industry, Harbin, China, 2016.
27.
YangFMaPLiW, et al. Intelligent ranking evaluation method of simulation models based on Siamese network. Syst Eng Electron2023; 45: 2060–2068.
28.
HamiltonWLYingRLeskovecJ.Representation learning on graphs: methods and applications. IEEE Data Eng Bullet2017; 40: 52–74.
29.
MuhammadBAQiCWuZ, et al. GRL-LS: a learning style detection in online education using graph representation learning. Expert Syst Appl2022; 201: 117138.
30.
LiMMHuangKZitnikM.Graph representation learning in biomedicine and healthcare. Nat Biomed Eng2022; 6: 1353–1369.
31.
ShahSMAHShahSFHHussainS, et al. Traffic forecasting route optimization in smart environment using graph representation learning. In: Proceedings of the 2023 IEEE international smart cities conference (ISC2), Bucharest, 24–27 September 2023, pp. 1–5. New York: IEEE.
32.
VeličkovićPFedusWHamiltonWL, et al. Deep graph infomax. arXiv preprint arXiv: 1809.10341, 2018.
33.
PengZHuangWLuoM, et al. Graph Representation learning via graphical mutual information maximization. In: Proceedings of the 29th international World Wide Web conference, Taipei, Taiwan, 20–24 April 2020, pp. 259–270. New York: ACM.
34.
SunFYHoffmannJVermaV, et al. InfoGraph: unsupervised and semi-supervised graph-level representation learning via mutual information maximization. arXiv preprint arXiv: 1908.01000, 2019.
35.
GoodfellowIBengioYCourvilleA.Deep learning. Cambridge, MA: The MIT Press, 2016, pp. 164–169.
SalehiADavulcuH.Graph attention auto-encoders. In: Proceedings of the 2020 IEEE 32nd international conference on tools with artificial intelligence (ICTAI), Baltimore, MD, 9–11 November 2020, pp. 989–996. New York: IEEE.
38.
HamiltonWLYingRLeskovecJ. Inductive representation learning on large graphs. In: Proceedings of the 31st international conference on neural information processing systems (NIPS 2017) (Advances in neural information processing systems), Long Beach, CA, 4–9 December 2017, pp. 1025–1035. Red Hook, NY: Curran Associates.
39.
XuKHuWLeskovecJ, et al. How powerful are graph neural networks? arXiv preprint arXiv: 1810.00826, 2018.
40.
SchlichtkrullMKipfTNBloemP, et al. Modeling relational data with Graph Convolutional Networks. In: Proceedings of the 15th international conference on semantic web, Heraklion, 3–7 June 2018, pp. 593–607. Berlin; Heidelberg: Springer-Verlag.
41.
DaiHDaiBSongL. Discriminative embeddings of latent variable models for structured data. In: Proceedings of the 33rd international conference on machine learning (ICML), New York, 19–24 June 2016, pp. 2702–2711New York: PMLR.
42.
DengAHooiB.Graph Neural Network-based anomaly detection in multivariate time series. In: Proceedings of the 35th AAAI conference on artificial intelligence (AAAI), 2–9 February 2021, pp. 4027–4035. Palo Alto, California: AAAI Press.
43.
SzekelyGJRizzoMLBakirovNK.Measuring and testing dependence by correlation of distance. Ann Stat2008; 35: 2769–2794.
44.
WestDB.Introduction to graph theory. 2nd ed.Hoboken, NJ: Prentice Hall, 2001, pp. 1–2.
45.
YingRYouJMorrisC, et al. Hierarchical graph representation learning with differentiable pooling. In: Proceedings of the 32nd international conference on neural information processing systems (NIPS 2018) (Advances in neural information processing systems), Montréal, QC, Canada, 3–8 December 2018, pp. 4800–4810. Red Hook, NY: Curran Associates.
46.
ZhangMCuiZNeumannM, et al. An end-to-end deep learning architecture for graph classification. In: Proceedings of the 32nd AAAI conference on artificial intelligence, New Orleans, LA, 2–7 February 2018, pp. 4438–4445. Palo Alto, California: AAAI Press.
47.
CaiCXuDFangC, et al. Graph Neural Networks for the cross-domain histopathological image classification. In: Proceedings of the 2021 IEEE 18th international symposium on biomedical imaging (ISBI), Nice, 13–16 April 2021, pp. 1953–1957. New York: IEEE.
48.
JaisIKMIsmailARNisaSQ. Adam optimization algorithm for wide and deep neural network. Knowl Eng Data Sci2019; 2: 41.
