A particle swarm optimization approach in printed circuit board thermal design

Printed circuit boards (PCBs) have dominated the electronics market, being part of nearly any commercial electronic device. However, the requirement for smaller and more efficient electronic equipment poses challenging problems in their thermal design. In this work we present a novel method for optimizing the design of PCBs in terms of thermal operation, based on comprehensive learning particle swarm optimization (CLPSO). Firstly, an encoding scheme is introduced for representing the potential placement of electronic components on the board as particles. A specially tailored CLPSO algorithm is then applied to optimize the components' positions with respect to the temperatures generated on the PCB; the latter are calculated using a detailed three-dimensional thermal model. The algorithm includes mechanisms for preventing components to overlap or to be placed out of the bounds of the board. The proposed approach is evaluated on case studies involving the optimization of PCBs with components of different sizes, heat dissipation levels and optimal operating temperatures; results show that the resulting placement helps to reduce the temperature profile on the board, a fact which is very important in terms of PCB performance and reliability. A comparison with alternative PCB thermal optimization techniques, highlights the superiority of the proposed method.