Multi-objective optimization approach to enhance the stencil printing quality

Abstract

Stencil printing process (SPP) is a key process in surface mount technology (SMT) production lines, which attributes to more than 60% of the defects in the assembly of the printed circuit boards (PCBs). This research integrates multi-objective optimization and data mining to enhance the stencil printing quality. Support vector regression (SVR) is used to model the relationships between the printing variables (speed, pressure, and separation speed) and the volume transfer efficiency (TE). Three new objective functions are studied and compared, which aim to maximize the printing process capability index Ĉpk. The multi-objective optimization problem is converted into a single objective using the ε-constraint approach. Evolutionary strategies (ES) with an adaptive penalty function is used to handle the constraints and solve the optimization model. The optimal solutions are retrieved for both printing directions (forward (F) and backward (B)) using two spec limits’ strategies (fixed and customized). The results show that the optimal solutions obtained from solving the optimization model with different objective functions behave differently. Moreover, the customized spec limits outperform the fixed limits strategy.