报告题目： Global Optimization of a 100-Variable Simulation-Based Black-box Design Problem

主讲人： 加拿大Simon Fraser University Gary Wang教授

时 间：3月12日下午3.00-4.30

地 点：机械学院531报告厅

欢迎广大师生参加！

Abstract:

Assembly quality of an auto-body is directly affected by how different sheet metal parts of the body are assembled.

This work focuses on choosing the best locations for the locators and clamps at each assembly station for a 10part subassembly.

As each part needs two locators and three clamps, when considering the positions in 2D, there are 10 variables for each part.

In total, the problem involves 100 variables with 2x10172 possible solutions.

The quality of the assembly is assessed with MonteCarlo simulation.

This problem is thus a typical high dimensional, expensive, blackbox (HEB) optimization problem.

Product Design and Optimization Lab (PDOL), sponsored by a worldleading automaker and Canadian Natural Science

and Engineering Research Council (NSERC),

has successfully developed methods and tools that well solved this problem with only 104 simulations (or function evaluations)

and improved the assembly quality by more than 60%.

This work represents the state-of-the-art in modern design optimization.