top of page


(CANCELLED) Machine-learning-based asymptotic homogenisation/localisation and design of spatially-varying lattice configurations


Professor Yichao Zhu

School of Mechanics and Aerospace Engineering

Dalian University of Technology


Date & Time

Friday, 16 February 2024

7:30 am


Tam Wing Fan Innovation Wing Two G/F, Run Run Shaw Building, HKU

Moderator:   Professor David Srolovitz


In this talk, we introduce a general framework, with the use of traditional asymptotic homogenisation approaches and the emerging machine learning tools, for the (CAD-inspired) representation, (CAE-inspired) analysis and (CAD-CAE integrated) design of smoothly-varying lattice configurations. Asymptotic analysis serves to identify the expressions for key quantities of interest, such as the (nonlinear) overall compliance, the sites and the magnitude of the maximum tensile stress, etc., in a scale-separated manner. Machine learning method is employed to embody those implicit interrelationships that are confirmed with asymptotic analysis. A number of simulation examples will be presented to show the balanced accuracy and efficiency of the proposed method. The presentation concludes with mathematical analogy to generalise the method used for analysing the behaviour of other multiscale systems.


[1] Zhu, Y. C., Li,S. S., Du, Z. L., Liu, C., Guo, X., and Zhang, W. S., “A novel asymptotic-analysis-based homogenisation approach towards fast design of infill graded microstructures”, J. Mech. Phys. Solids, 124, 612-633 (2019).

[2] Xue, D. C., Zhu, Y. C., and Guo, X., “Generation of smoothly-varying infill configurations from a continuous menu of cell patterns and the asymptotic analysis of its mechanical behaviour”, Comput. Methods Appl. Mech. Engrg., 366, 113037 (2020).

[3] Li, S. S., Zhu, Y. C., and Guo, X., “Optimisation of spatially varying orthotropic porous structures based on conformal mapping”, Comput. Methods Appl. Mech. Engrg., 391, 114589 (2022).

[4] Ma, C., Xue, D. C., Li, S. S., Zhou, Z. C., Zhu, Y. C., and Guo, X., “Compliance minimisation of smoothly varying multiscale structures using asymptotic analysis and machine learning”, Comput. Methods Appl. Mech. Engrg., 395, 114861 (2022).

[5] Ma, C., Zhang, J. H., and Zhu, Y. C., “Performance analysis and optimisation of spatially-varying infill microstructure within CAD geometries”, Comput. Methods Appl. Mech. Engrg., 416, 116373 (2023).

[6] Zhou, Z. C., Zhu, Y. C., and X. Guo, X., “Machine learning based asymptotic homogenization and localization: Predictions of key local behaviors of multiscale configurations bearing microstructural varieties”, Int. J. Numer. Meth. Eng., 124(3), 639-669 (2023).

[7] Zhao, X., and Zhu, Y. C., “A general leading-order asymptotic theory of thin microstructural plates and uncertainty quantification of the elastic performance of composite laminates”, J. Comp. Mater., 57(20), 3145-3171 (2023).

[8] Pan, X. W., Zhou, Z. C., Ma, C., Li, S. S., and Zhu, Y. C., “Machine-learning-based asymptotic homogenisation and localisation considering boundary layer effects”, Int. J. Numer. Meth. Eng., 125, e7367 (2024).


Dr. Yichao Zhu is now a full professor based in the School of Mechanics and Aerospace Engineering at Dalian University of Technology (DUT). Dr. Zhu got his Bachelor degree on applied mathematics from Fudan University, and his PhD on applied mathematics from University of Oxford. Dr. Zhu’s major research interest lies in the modelling and simulations of multiple-scale problems that arise in physics, engineering and materials science, and he is now devoting efforts to re-vitalising traditional asymptotic homogenisation/localisation techniques with the use of machine learning. Dr. Zhu also bears strong interests in projects, where mathematical tools can be used by all means to solve industrial problems. Dr. Zhu has published more than 40 peer-reviewed articles. He sits in the junior editorial board of Acta Mechanica Solidia Sinica, and serves in the administrative panel on mechanics of soft materials underneath the Chinese Society of Theoretical and Applied Mechanics.


Research Areas:

Contact for


Download Details

in PDF

bottom of page