Multi-Objective Optimization Design of Passive Compliant Wrist Structure of a Manipulator Based on RBF Neural Network Model
Machines · 07.11.2026 · 14 (11) · 1315
DOI: 10.3390/machines14111315 ↗Kurz gefasst
Fachartikel in der Zeitschrift Machines, online erschienen am 07.11.2026. Verfasst von Dongxia Wang und 6 weiteren Autor:innen. Die Arbeit nennt 29 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Henan Science and Technology Department.
Abstract (Originalsprache)
Passive compliant wrists compensate for assembly misalignment through elastic deformation, making them suitable for handling thin-walled brittle parts. However, the trade-offs among compensation capacity, dynamic performance, and stress remain insufficiently characterized for the spatial variable-section wrist considered in this study. A data-driven multi-objective optimization framework is developed to address these trade-offs using deformation compensation, first-order natural frequency, and maximum equivalent stress as performance indicators. A Plackett–Burman design identifies the key geometric parameters, and optimal Latin hypercube sampling provides training data for a radial basis function (RBF) neural network surrogate. The surrogate is coupled with the non-dominated sorting genetic algorithm II (NSGA-II) to generate a set of non-dominated designs. An entropy-weighted technique for order preference by similarity to the ideal solution (TOPSIS) selects a representative compromise, reducing reliance on manually assigned objective weights. Finite-element verification yields a deformation compensation of 0.51 mm, a first-order natural frequency of 137.78 Hz, and a maximum equivalent stress of 39.05 MPa. Compared with the baseline design, deformation compensation increases by 30.77% and maximum equivalent stress decreases by 13.74%, with a 2.81% reduction in first-order natural frequency. Prototype assembly tests under prescribed angular and lateral misalignments further demonstrate the error-compensation capability of the optimized wrist.
Förderung
- National Natural Science Foundation of China
- Henan Science and Technology Department
Zitieren
APA
Wang, D., Yao, W., Wang, L., Wang, C., Zhang, Z., Ding, S., & Liu, K. (2026). Multi-Objective Optimization Design of Passive Compliant Wrist Structure of a Manipulator Based on RBF Neural Network Model. Machines, 14(11), 1315. https://doi.org/10.3390/machines14111315
Vancouver
Wang D, Yao W, Wang L, Wang C, Zhang Z, Ding S, et al. Multi-Objective Optimization Design of Passive Compliant Wrist Structure of a Manipulator Based on RBF Neural Network Model. Machines. 2026;14(11):1315. doi:10.3390/machines14111315
BibTeX
@article{Wang2026,
author = {Wang, Dongxia and Yao, Wang and Wang, Liangwen and Wang, Caidong and Zhang, Zhichong and Ding, Shixing and Liu, Kangzhe},
title = {Multi-Objective Optimization Design of Passive Compliant Wrist Structure of a Manipulator Based on RBF Neural Network Model},
journal = {Machines},
year = {2026},
volume = {14},
number = {11},
pages = {1315},
doi = {10.3390/machines14111315}
}
EUVIMED