A genetic algorithm based solution for the online cutting problem with defects

Abstract

In flat glass production, glass products of various sizes and quality classes are cut from a continuously moving glass ribbon. If a product is placed on a part of the glass ribbon that has more defects than allowed by that product's quality class, it is discarded as scrap. Scrap glass is sent back to the furnace to be recycled. This process creates additional production costs and reduces the productivity of the line. Therefore, the main objective of the cutting process is the minimization of scrap glass. In this study, we propose an online solution approach that uses a look-ahead mechanism to make cutting decisions in real time. This approach solves a series of static cutting problems over a rolling horizon using a genetic algorithm. The proposed approach has been tested successfully on realistic problem instances that are based on parameters of contemporary glass production lines.