Multidisciplinary Optimization of Auto-Body Lightweight Design Using Hybrid Metamodeling Technique and Particle Swarm Optimizer

Because of rising complexity during the automotive product development process, the number of disciplines to be concerned has been significantly increased. Multidisciplinary design optimization (MDO) methodology, which provides an opportunity to integrate each discipline and conduct compromise searching process, is investigated and introduced to achieve the best compromise solution for the automotive industry. To make a better application of MDO, the suitable coupling strategy of different disciplines and efficient optimization techniques for automotive design are studied in this article. Firstly, considering the characteristics of automotive load cases which include many shared variables but rare coupling variables, a multilevel MDO coupling strategy based on enhanced collaborative optimization (ECO) is studied to improve the computational efficiency of MDO problems. Then, a hybrid metamodeling technique is developed to surrogate the time-consuming simulation analysis with local and global metamodels, aiming at balancing accuracy and efficiency of metamodel construction process. At last, the particle swarm optimizer is employed and adjusted to combine with the constructed hybrid metamodels for conducting the optimization program of the MDO problems. In order to improve the global optimizing capability of particle swarm optimization (PSO) algorithm, the diversity-enhanced mechanism and local search method are used to modify the searching process. The established MDO architecture is applied to a lightweight design application of an auto-body, and the results verify its effectiveness and validity.