An efficient genetic algorithm for setup time minimization in PCB assembly

Abstract

A central aspect of surface mount technology (SMT) systems is the assembly of printed circuit boards (PCBs) which requires the resolution of many optimization problems. One of these problems arises when assembling many types of PCBs on a single machine. In this case, the main goal becomes the minimization of the setup times. That is, the time required to modify the feeder rack in order to have all components needed by the next type of PCB to be assembled. Achieving such a minimization goal will provide the system with improved productivity and flexibility capabilities. In order to minimize the setup time, we propose a genetic algorithm that uses a group-based representation with a series of specialized genetic operators. A set of 90 instances is proposed as a test bed for the single machine many-types of PCB problem. The proposed algorithm accomplishes the best known result for a benchmark instance of the problem and outperforms, in terms of assembly time, a well known heuristic on the set of proposed instances.