Title: Quality design for laser micro-manufacturing processes using Bayesian modeling and optimization methods
Authors: Jianjun Wang - Nanjing University of Science and Technology (China) [presenting]
Abstract: The micro-manufacturing process usually has some typical features such as multiple noise factors, large variations, high manufacturing costs and poor repeatability. In view of high uncertainties for quality design problems in the micro-manufacturing process, a unified framework of Bayesian modeling and optimization is proposed to improve product quality and reduce manufacturing cost. First of all, Bayesian modeling methods are utilized to develop the relationship models between input factors (e.g., laser average power, pulse frequency and cutting speed) and output quality characteristics (e.g., hole diameter and roundness) in the laser micro-drilling process. Then, the simulated responses which reflect a real laser micro-drilling process are obtained by using a Gibbs sampling procedure. Furthermore, the cost structures for mixed multiple quality characteristics is analyzed and then the rejection cost (i.e., rework cost and scrap cost) function is constructed by using the simulated response values. Finally, the optimum economic parameter settings of laser micro-drilling process can be obtained by optimizing the proposed cost function with a hybrid genetic algorithm. The results show the proposed approach can significantly improve the product quality and reduce the rejection cost in the micro-drilling process.