EUVIMEDEUVIMED entdecken
Crossref

Topology-aware opto-mechanical optimization of mirror support structures via a hybrid GA-CNN-LSTM surrogate model

Yuan Chen, Jin-Song Wang, Hao Wang

Measurement Science and Technology · 26.06.2026 · 37 (26) · 265201

DOI: 10.1088/1361-6501/ae7c8f ↗

Kurz gefasst

Fachartikel in der Zeitschrift Measurement Science and Technology, online erschienen am 30.06.2026. Verfasst von Yuan Chen und 2 weiteren Autor:innen. Die Arbeit nennt 29 Literaturangaben. Gefördert von: National Key Project of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The design of high-performance mirror support structures entails navigating a complex parameter space where optical surface accuracy inherently conflicts with mechanical stiffness. Conventional finite element analysis-driven optimization is computationally prohibitive and prone to local convergence when handling high-dimensional nonlinear couplings. To address these bottlenecks, this paper proposes a hierarchical optimization framework driven by a topology-aware genetic algorithm (GA)-convolutional neural networks (CNNs)-long short-term memory (LSTM) surrogate model. In this architecture, CNN are employed to encode the spatial topology of the support structure, while LSTM networks capture the sequential dependencies of the structural load path. A GA is further used to adaptively optimize the network hyperparameters, thereby improving predictive fidelity for opto-mechanical responses. By embedding the trained surrogate into the non-dominated sorting GA II algorithm, the framework enables rapid global multi-objective optimization. The optimized design achieves a surface figure RMS of 18.1 nm, a fundamental frequency of 86.4 Hz, and a maximum equivalent stress of 0.61 MPa. Experimental interferometric validation further yields a measured RMS of 22.24 nm, confirming the practical feasibility of the proposed framework. These results demonstrate that the proposed method effectively balances design precision and computational efficiency, providing a robust solution for complex opto-mechanical engineering problems.

Förderung

  • National Key Project of China

Zitieren

APA

Chen, Y., Wang, J., & Wang, H. (2026). Topology-aware opto-mechanical optimization of mirror support structures via a hybrid GA-CNN-LSTM surrogate model. Measurement Science and Technology, 37(26), 265201. https://doi.org/10.1088/1361-6501/ae7c8f

Vancouver

Chen Y, Wang J, Wang H. Topology-aware opto-mechanical optimization of mirror support structures via a hybrid GA-CNN-LSTM surrogate model. Measurement Science and Technology. 2026;37(26):265201. doi:10.1088/1361-6501/ae7c8f

BibTeX
@article{Chen2026,
  author = {Chen, Yuan and Wang, Jin-Song and Wang, Hao},
  title = {Topology-aware opto-mechanical optimization of mirror support structures via a hybrid GA-CNN-LSTM surrogate model},
  journal = {Measurement Science and Technology},
  year = {2026},
  volume = {37},
  number = {26},
  pages = {265201},
  doi = {10.1088/1361-6501/ae7c8f}
}
Kostenlos starten →