Dazhi Zhao 赵大志
Hi! I am a third-year undergraduate majoring in Engineering Mechanics at the School of Aerospace Engineering and Applied Mechanics, Tongji University. I was born in Xi'an, one of China's historic capitals.
My research interests include: Data-driven design Differentiable simulation Inverse problem
I am advised by Prof. Keke Tang of FCM Lab and also by Prof. Rui Fan of the MIAS Group, both at Tongji University.
News
2026/04/18
Our Composite Structures paper, Physics-constrained neural networks for high-fidelity composite failure envelopes, is now online.
2026/03/17
Our preprint on physics-guided diffusion models for inverse design of disordered metamaterials is now available on arXiv.
2026/03/08
Our Thin-Walled Structures paper on autoregressive inverse design of disordered metamaterials was accepted.
2026/01/28
Started my internship at HKUST and began collaborating with Prof. Tianju Xue on differentiable simulation.
2025/07/18
Presented our work on rapid prediction and impact-parameter identification of interlaminar damage at ICDM 2025 in Singapore.
Publications
Autoregressive Inverse Design of Disordered Metamaterials for Target Nonlinear Response
Zhao, D., Xiang, Y., Zhang, P., Liu, N., Wang, X., and Tang, K. (2026). "Autoregressive inverse design of disordered metamaterials for target nonlinear response." Thin-Walled Structures, 225, 114793.
Physics-Constrained Neural Networks for High-Fidelity Composite Failure Envelopes
Zhang, R., Zhao, D., Zhang, P., and Tang, K. (2026). "Physics-constrained neural networks for high-fidelity composite failure envelopes." Composite Structures, 120358.
Physics-Guided Diffusion Models for Inverse Design of Disordered Metamaterials
Xie, Z., Xu, W., Zhao, D., Zhang, W., Dong, D., Xu, B., Liu, N., Mao, S., and Xue, T. (2026). "Physics-guided diffusion models for inverse design of disordered metamaterials." arXiv preprint arXiv:2603.16209.
Projects
GUI for Phase-Field Fracture Workflows
An interface-oriented research tool that streamlines phase-field fracture simulation workflows, making model setup, execution, and result inspection more accessible.
Reinforcement Learning for Torque Control in a Four-Link Mechanism
A control-oriented project that explores how reinforcement learning can generate effective torque strategies for a coupled four-link mechanical system.