Component grouping for printed circuit board assembly

Abstract

In a Printed Circuit Board (PCB) assembly company, an assembly line is designed to have several placement machines (may or may not be the same types) to deal with various types of components since the production volume is large and the placement operation is a bottleneck of the line. This research project studies the component grouping problem in order to assign components to proper placement machines so that the optimal cycle time and throughput of the line can be obtained. In addition to an extensive literature review on PCB assembly being carried out, the component grouping problem was identified in the project. Several mathematical models were formulated to cope with different objectives for both single-sided and double-sided board assembly cases. Furthermore, a new linear programming algorithm was proposed to solve the component grouping problem with the objective of minimizing cycle time, based on the revised simplex method. Moreover, the Genetic Algorithm (GA) technique was applied to the component grouping problem, and the results from both the linear programming method and the GA technique were compared with the solution from the integer programming model, obtained by the commercial package, CPLEX. The conclusion is that the result generated by the linear programming method with rounding off is acceptable in engineering, whereas the solution improvement rate of the genetic algorithm is very good at the beginning, but it decreases quickly in an exponential rate as the solution gets closer to its optimal.