Unstructured surface mesh smoothing method based on deep reinforcement learning. (English) Zbl 1540.76176
Summary: In numerical simulations such as computational fluid dynamics simulations or finite element analyses, mesh quality affects simulation accuracy directly and significantly. Smoothing is one of the most widely adopted methods to improve unstructured mesh quality in mesh generation practices. Compared with the optimization-based smoothing method, heuristic smoothing methods are efficient but yield lower mesh quality. The balance between smoothing efficiency and mesh quality has been pursued in previous studies. In this paper, we propose a new smoothing method that combines the advantages of the heuristic Laplacian method and the optimization-based method based on the deep reinforcement learning method under the Deep Deterministic Policy Gradient framework. Within the framework, the actor artificial neural network predicts the optimal position of each interior free node with its surrounding ring nodes. At the same time, a critic-network is established and takes the mesh quality as input and outputs the reward of the action taken by the actor-network. Training of the networks will maximize the cumulative long-term reward, which ends up maximizing the mesh quality. Training and validation of the proposed method are presented both on 2-dimensional triangular meshes and 3-dimensional surface meshes, which demonstrates the efficiency and mesh quality of the proposed method. Finally, numerical simulations on perturbed meshes and smoothed meshes are carried out and compared which prove the influence of mesh quality on the simulation accuracy.
MSC:
| 76M99 | Basic methods in fluid mechanics |
| 76G25 | General aerodynamics and subsonic flows |
| 76H05 | Transonic flows |
| 68T05 | Learning and adaptive systems in artificial intelligence |
| 68T07 | Artificial neural networks and deep learning |
Keywords:
topological mesh optimization; artificial neural network; Laplacian smoothing; optimization-based smoothing; subsonic flow; NACA0012 airfoil; transonic flowReferences:
| [1] | Mavriplis, DJ, Unstructured grid techniques, Annu Rev Fluid Mech, 29, 1, 473-514, (1997) · doi:10.1146/annurev.fluid.29.1.473 |
| [2] | Baker, TJ, Mesh generation: Art or science?, Prog Aerosp Sci, 41, 1, 29-63, (2005) · doi:10.1016/j.paerosci.2005.02.002 |
| [3] | Liu, WK; Li, SF; Park, HS, Eighty years of the finite element method: birth, evolution, and future, Arch Comput Methods Eng, 29, 4431-4453, (2022) · doi:10.1007/s11831-022-09740-9 |
| [4] | Katz, A.; Sankaran, V., Mesh quality effects on the accuracy of CFD solutions on unstructured meshes, J Comput Phys, 230, 20, 7670-7686, (2011) · Zbl 1452.76198 · doi:10.1016/j.jcp.2011.06.023 |
| [5] | Katz, A.; Sankaran, V., High aspect ratio grid effects on the accuracy of Navier-Stokes solutions on unstructured meshes, Comput Fluids, 65, 66-79, (2012) · Zbl 1365.76194 · doi:10.1016/j.compfluid.2012.02.012 |
| [6] | Wang, N.; Li, M.; Ma, R., Accuracy analysis of gradient reconstruction on isotropic unstructured meshes and its effects on inviscid flow simulation, Adv Aerodyn, 1, 1, 1-31, (2019) · doi:10.1186/s42774-019-0020-9 |
| [7] | Ho-Le, K., Finite element mesh generation methods: a review and classification, Comput Aided Des, 20, 1, 27-38, (1988) · Zbl 0661.65124 · doi:10.1016/0010-4485(88)90138-8 |
| [8] | Weatherill, NP; Hassan, O., Efficient three-dimensional grid generation using the Delaunay triangulation, Comput Fluid Dyn, 1992, 961-968, (1992) |
| [9] | Du, Q.; Wang, D., Recent progress in robust and quality Delaunay mesh generation, J Comput Appl Math, 195, 1-2, 8-23, (2006) · Zbl 1096.65016 · doi:10.1016/j.cam.2005.07.014 |
| [10] | Turk G (1992) Re-tiling polygonal surfaces. In: Proceedings of the 19th annual conference on Computer graphics and interactive techniques, pp 55-64 |
| [11] | Liu, J.; Chen, YQ; Sun, SL, Small polyhedron reconnection for mesh improvement and its implementation based on advancing front technique, Int J Numer Meth Eng, 79, 8, 1004-1018, (2009) · Zbl 1171.74472 · doi:10.1002/nme.2605 |
| [12] | Gonzaga de Oliveira SL (2012) A review on Delaunay refinement techniques. International Conference on Computational Science and Its Applications, Springer, Berlin, Heidelberg, pp 172-187 |
| [13] | Field, DA, Laplacian smoothing and Delaunay triangulations, Commun Appl Numer Methods, 4, 6, 709-712, (1988) · Zbl 0664.65107 · doi:10.1002/cnm.1630040603 |
| [14] | Vartziotis, D.; Athanasiadis, T.; Goudas, I., Mesh smoothing using the geometric element transformation method, Comput Methods Appl Mech Eng, 197, 45-48, 3760-3767, (2008) · Zbl 1197.65197 · doi:10.1016/j.cma.2008.02.028 |
| [15] | Zhou T, Shimada K (2000) An angle-based approach to two-dimensional mesh smoothing. in: Proceedings 9th International Meshing Roundtable 373-384 |
| [16] | Vartziotis, D.; Wipper, J., The geometric element transformation method for mixed mesh smoothing, Eng Comput, 25, 3, 287-301, (2009) · doi:10.1007/s00366-009-0125-6 |
| [17] | Vartziotis, D.; Wipper, J., Fast smoothing of mixed volume meshes based on the effective geometric element transformation method, Comput Methods Appl Mech Eng, 201, 65-81, (2012) · Zbl 1239.65075 · doi:10.1016/j.cma.2011.09.008 |
| [18] | Freitag L, Knupp P, Munson T et al (2002) A comparison of optimization software for mesh shape-quality improvement problems. in: Proceedings 11th International Meshing Roundtable 29-40 |
| [19] | Garimella RV, Shashkov MJ, Knupp PM (2002) Optimization of surface mesh quality using local parametrization. in: Proceedings 11th International Meshing Roundtable 41-52 |
| [20] | Brewer M, Freitag-Diachin L, Knupp P et al (2003) The mesquite mesh quality improvement toolkit. in: Proceedings 12th International Meshing Roundtable 239-250 |
| [21] | Raissi, M.; Perdikaris, P.; Karniadakis, GE, Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations, J Comput Phys, 378, 686-707, (2019) · Zbl 1415.68175 · doi:10.1016/j.jcp.2018.10.045 |
| [22] | Sirignano, J.; Spiliopoulos, K., DGM: a deep learning algorithm for solving partial differential equations, J Comput Phys, 375, 1339-1364, (2018) · Zbl 1416.65394 · doi:10.1016/j.jcp.2018.08.029 |
| [23] | Haghighat, E., A physics-informed deep learning framework for inversion and surrogate modeling in solid mechanics, Comput Methods Appl Mech Eng, 379, 113741, (2021) · Zbl 1506.74476 · doi:10.1016/j.cma.2021.113741 |
| [24] | Zhu, Y.; Zabaras, N.; Koutsourelakis, PS, Physics-constrained deep learning for high-dimensional surrogate modeling and uncertainty quantification without labeled data, J Comput Phys, 394, 56-81, (2019) · Zbl 1452.68172 · doi:10.1016/j.jcp.2019.05.024 |
| [25] | Brunton, SL; Noack, BR; Koumoutsakos, P., Machine learning for fluid mechanics, Annu Rev Fluid Mech, 52, 1, 477-508, (2020) · Zbl 1439.76138 · doi:10.1146/annurev-fluid-010719-060214 |
| [26] | Singh, AP; Medida, S.; Duraisamy, K., Machine learning-augmented predictive modeling of turbulent separated flows over airfoils, AIAA J, 55, 7, 2215-2227, (2017) · doi:10.2514/1.J055595 |
| [27] | Bhatnagar, S.; Afshar, Y.; Pan, S.; Duraisamy, K.; Kaushik, S., Prediction of aerodynamic flow fields using convolutional neural networks, Comput Mech, 64, 2, 525-545, (2019) · Zbl 1468.76051 · doi:10.1007/s00466-019-01740-0 |
| [28] | Saha, S.; Gan, Z.; Cheng, L., Hierarchical deep learning neural network (HiDeNN): an artificial intelligence (AI) framework for computational science and engineering, Comput Methods Appl Mech Eng, 373, 113452, (2021) · Zbl 1506.68110 · doi:10.1016/j.cma.2020.113452 |
| [29] | Zhang, L.; Cheng, L.; Li, H., Hierarchical deep-learning neural networks: finite elements and beyond, Comput Mech, 67, 1, 207-230, (2021) · Zbl 07360501 · doi:10.1007/s00466-020-01928-9 |
| [30] | Wu, TF; Liu, X.; Wei, An, A mesh optimization method using machine learning technique and variational mesh adaptation, Chin J Aeronaut, 35, 3, 27-41, (2022) · doi:10.1016/j.cja.2021.05.018 |
| [31] | Jiang M, Gallagher B, Mandell N et al (2019) A deep learning framework for mesh relaxation in arbitrary Lagrangian-Eulerian simulations. Proceedings of SPIE 111390O |
| [32] | Fidkowski, KJ; Chen, G., Metric-based, goal-oriented mesh adaptation using machine learning, J Comput Phys, 426, 109957, (2020) · Zbl 07510070 · doi:10.1016/j.jcp.2020.109957 |
| [33] | Wang, NH; Lu, P.; Chang, XH, Preliminary investigation on unstructured mesh generation technique based on advancing front method and machine learning methods, Chin J Theoret Appl Mech, 53, 3, 740-751, (2021) |
| [34] | Wang, NH; Lu, P.; Chang, XH, Unstructured mesh size control method based on artificial neural network, Chin J Theoret Appl Mech, 53, 10, 2682-2691, (2021) |
| [35] | Chen, X.; Li, T.; Wan, Q., MGNet: a novel differential mesh generation method based on unsupervised neural networks, Eng Comput, 38, 4409-4421, (2022) · doi:10.1007/s00366-022-01632-7 |
| [36] | Pan, J.; Huang, J.; Wang, Y., A self-learning finite element extraction system based on reinforcement learning, Artif Intell Eng Des Anal Manuf, 35, 2, 1-29, (2021) · doi:10.1017/S089006042100007X |
| [37] | Pan J, Huang J, Cheng G et al (2022) Reinforcement learning for automatic quadrilateral mesh generation: a soft actor-critic approach. arXiv preprint: 2203.11203 |
| [38] | Guo, YF; Wang, CR; Ma, Z., A new mesh smoothing method based on a neural network, Comput Mech, 69, 425-438, (2022) · Zbl 07492678 · doi:10.1007/s00466-021-02097-z |
| [39] | Lillicrap P, Hunt JJ, Pritzel A et al (2015) Continuous control with deep reinforcement learning. conference paper, ICLR2016, arXiv preprint: 1509.02971 |
| [40] | Digital shape workbench v5.0 AIM@SHAPE http://visionair.ge.imati.cnr.it/ontologies/shapes/ |
| [41] | He, X.; Zhao, Z.; Ma, R., Validation of HyperFLOW in subsonic and transonic flow, Acta Aerodyn Sin, 34, 2, 267-275, (2016) |
| [42] | He, X.; He, XY; He, L., HyperFLOW: a structured/unstructured hybrid integrated computational environment for multi-purpose fluid simulation, Proced Eng, 126, 645-649, (2015) · doi:10.1016/j.proeng.2015.11.254 |
| [43] | Wang, NH; Li, M.; Zhang, LP, Accuracy analysis and improvement of viscous flux schemes in unstructured second-order finite-volume discretization, Chin J Theoret Appl Mech, 50, 3, 527-537, (2018) |
| [44] | Fidkowski, KJ; Darmofal, DL, Review of output-based error estimation and mesh adaptation in computational fluid dynamics, AIAA J, 49, 4, 673-694, (2011) · doi:10.2514/1.J050073 |
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.
