Research

Computational Intelligence

“Use computers to think, learn and create”

Research Topic

Artificial Intelligence & Machine Learning

We develop AI/ML methods to facilitate the representation, evaluation, knowledge discovery, and inverse design of physical intelligence systems. We emphasize the integration of physical knowledge, interpretability, and uncertainty quantification within the machine learning models. 

  • Database generation and quality control
  • Gaussain process and active learning
  • Deep generative modeling
  • Physics-informed machine learning
  • Multi-source/fidelity data fusion

 

  1. Wang, L., Chan, Y.C., Ahmed, F., Liu, Z., Zhu, P. and Chen, W., 2020. Deep generative modeling for mechanistic-based learning and design of metamaterial systems. Computer Methods in Applied Mechanics and Engineering, 372, p.113377.
  2. Wang, L., Yerramilli, S., Iyer, A., Apley, D., Zhu, P. and Chen, W., 2022. Scalable gaussian processes for data-driven design using big data with categorical factors. Journal of Mechanical Design, 144(2), p.021703.
  3. Chen, Y.P., Wang, L., Comlek, Y. and Chen, W., 2024. A Latent Variable Approach for Non-Hierarchical Multi-Fidelity Adaptive Sampling. Computer Methods in Applied Mechanics and Engineering, 421, p.116773.
  4. Lee, D., Chan, Y.C., Chen, W., Wang, L., van Beek, A. and Chen, W., 2023. t-metaset: Task-aware acquisition of metamaterial datasets through diversity-based active learning. Journal of Mechanical Design, 145(3), p.031704.
Research Topic

Design & Optimization

Beyond forward modeling, we integrate AI/ML with optimization methods to lay down inverse design principles and frameworks for physical intelligence. We aim to tackle prevalent design challenges such as combinatorial space, entangled entities, complex physics, competing objectives, manufacturing constraints and uncertainty.

  • Topology optimization 
  • Multi-scale/physics optimization
  • Co-design methods (materials+structure+stimuli+process)
  • Bayesian optimization
  • Design under uncertainty

 

  1. Wang, L., Boddapati, J., Liu, K., Zhu, P., Daraio, C. and Chen, W., 2022. Mechanical cloak via data-driven aperiodic metamaterial design. Proceedings of the National Academy of Sciences, 119(13), p.e2122185119.
  2. Wang, L., Chang, Y., Wu, S., Zhao, R.R. and Chen, W., 2023. Physics-aware differentiable design of magnetically actuated kirigami for shape morphing. Nature Communications, 14(1), p.8516.
  3. Wang, L., Tao, S., Zhu, P. and Chen, W., 2021. Data-driven topology optimization with multiclass microstructures using latent variable Gaussian process. Journal of Mechanical Design, 143(3), p.031708.
  4. Wang, L., Liu, Z., Da, D., Chan, Y.C., Chen, W. and Zhu, P., 2022. Generalized de-homogenization via sawtooth-function-based mapping and its demonstration on data-driven frequency response optimization. Computer Methods in Applied Mechanics and Engineering, 395, p.114967.
Research Topic

Computational Simulation

We create scalable, efficient and differentiable simulation models for physical intelligence systems. We ultilize them to support data generation and gradient-based optimization. They are integrated with AI/ML to amplify individual capabilities, tackling challenges that surpass the scope of AI/ML alone.

 

  • Differentiable simulation
  • Reduce order modeling
  • Finite element method 

 

  1. Wang, L., Chang, Y., Wu, S., Zhao, R.R. and Chen, W., 2023. Physics-aware differentiable design of magnetically actuated kirigami for shape morphing. Nature Communications, 14(1), p.8516.
  2. Xu, W., Wang, L., Liu, Z. and Zhu, P., 2023. General assembly rules for metamaterials with scalable twist effects. International Journal of Mechanical Sciences, 259, p.108579.