Metamodeling approach in solving the machine parameters optimization problem using neural network and genetic algorithms: A case study

The use of multilayer ceramic capacitors (MLCCs) is increasing because they are surface-mountable and are used primarily in the expanding communication and computing market. In the MLCC manufacturing process, some 80% of the loss in yield is attributable to paste-printing quality problems. Improvement in the quality of MLCC screen-printing is therefore tactically and strategically important. This research extends existing MLCC screen-printing robust design results to search for a universal optimum solution. A metamodeling approach has been applied to solving a variety of optimization problems. This is an abstraction model form from a model. The abstracted model aims to reduce model complexity, and yet maintain validity. This work involved building a screen-printing quality metamodel, based upon fractional factorial experimental design data using a neural network approach—that were then solved by genetic algorithms. The empirical results are promising. The paper concludes with practical constraints and insights for management.