A path planning algorithm for PCB surface quality automatic inspection

The surface quality inspection of industrial printed circuit board (PCB) is a vitally important link in its manufacturing process. To inspect surface defects of PCBs effectively, the automatic optical inspection (AOI) technology, in which the PCB image acquisition depends on the path planning method, is widely adopted by industry. It is regarded as a characteristic travelling salesman problem (TSP), which includes component clustering, location adjustment and algorithm adaptation optimization. In this paper, by improving the ant colony algorithm (ACA) algorithm, we devise a PCB image acquisition path planning model and the corresponding solving algorithms. Because the ACA encounters difficulty escaping from the local optimal solution, an improved ACA with a negative feedback mechanism is proposed that is able to obtain a better tour path with a higher probability. Aiming at the uncertainty of the local location of image acquisition windows, location adjustment methods are introduced to further shorten the path length and improve the image acquisition efficiency. Finally, via simulation experiments, the proposed global negative feedback ACA (GNF-ACA) can shorten the average length of the tour path by 1.7% without changing the time complexity. The three methods of location adjustment can further shorten the length of the tour path by 5.6%, 13.1% and 13.7%.