An Object-Based Evolutionary Algorithm: The Nesting Solution
This document has been relocated to http://scholarsmine.mst.edu/engman_syseng_facwork/244
There were 12 downloads as of 28 Jun 2016.
Abstract
The nesting problems have received considerable attention and have been addressed by a variety of algorithms. Recently, evolutionary algorithms have been adopted for solutions. Most of these algorithms, however, require a search in one-dimensional space; thus a transformation of the problem to a single dimension, as in the sequencing problems, is needed. Unfortunately, this restricts the search space. In this study an object-based evolutionary algorithm for the nesting problems is proposed. The methodology is created in a true two-dimensional space, allowing object-based mechanisms and object-based evolutionary operators to perform effectively on the space without restricting search alternatives. Implementation of the algorithm is conducted using grid representation where no overlapping is allowed. Layout simulation/animation over generations shows the continual improvement by this method. Experimental results of packing density on rectangular and irregular versions of the nesting problem are up to 94.41% and 82.34%, respectively. For industrial-size data, five hundred and forty-three pieces are tested. The final packing density is 74.89%
