.png){kind=icon}
.png){kind=icon}
Return
Double ensemble model based RBDO method using ILCVT sampling
.png){kind=icon}
PRE
AI
DOI：10.1007/s12206-025-0948-0.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
To ease the choosing difficulty of RBF, Kriging and corresponding basis/correlation functions for unknown implicit problem in RBDO, a novel double ensemble (DE) modelling strategy was developed. Firstly, different basis/correlation functions of RBF and Kriging were weighted to obtain the ensemble RBF and ensemble Kriging models, which were further weighted to obtain the DE model. A heuristic weight calculation method was used to calculate weight coefficients of each sub surrogate and sub-ensemble model. To generate samples evenly and ensure the accuracy of the DE model around the initial design point, the inherited latinized centroidal Voronoi tessellation (ILCVT) sampling method was developed. Finally, the lightweight design of the cutterhead was performed by calling the DE model. Monte Carlo simulation and sequential quadratic programming were used to calculate the failure probability and iterative design point. After optimization, the mass of the cutterhead was reduced by 13.8 %, demonstrating the feasibility of proposed method.
Keywords:
Double ensemble model
Surrogate model
Optimization
ILCVT sampling
Journal
IF:
1.7
Papers:
601
Citations:
1.2W