This study aims to address the inefficiency of conventional multidisciplinary design optimization (MDO) methods in handling the implicit relationship between structural parameters and performance responses, as well as the complex coupling relationships among multiple performance responses during intelligent large-scale equipment optimization.
An improved collaborative optimization (CO) framework is developed through two key innovations. (1) An ensemble surrogate model combining radial basis function (RBF), Kriging and support vector regression (SVR) models is developed through linear weighting for enhanced approximation accuracy. (2) A modified CO algorithm incorporating dynamic penalty functions (ICO-DP) is proposed to effectively manage multiple coupling performance responses like mass minimization, fatigue life maximization and first-order modal frequency maximization. The methodology is validated using an earth pressure balance shield machine cutterhead as an engineering case study.
The optimization results show that the proposed ICO-DP method based on an ensemble surrogate model has obvious advantages over traditional methods, the mass is reduced by 3.2%, the first mode frequency is increased by 21.9% and the fatigue life is increased by 90%, which significantly improves the performance of the cutterhead.
This work makes three original contributions: (1) Ensemble surrogate model is applied to the design optimization of the cutterhead. (2) Integration of dynamic penalty functions with CO framework is conducted for coupled performance handling. (3) A complete methodological framework is built that bridges the gap between theoretical MDO research and practical large-scale equipment design applications.
