A hybrid approach for packing irregular patterns using evolutionary strategies and neural network

Packing problems can be found in various industries. Packing regular shapes (patterns) is common in wood, glass, and paper industries while packing irregular shapes can be found in metal, clothing and leather industries. It is obvious that irregular objects packing problems are more complex than regular ones due to the geometrical complexity of irregular shapes. Relatively few scientific approaches have been developed to solve irregular objects packing problems although the effectiveness of the packing approaches determines the industrial resource utilisation. This study constructs an irregular object packing approach that integrates a grid approximation-based representation, a learning vector quantisation neural network (NN), a heuristic placement strategy and an integer representation-based [image omitted] evolutionary strategy (ES) to obtain an efficient placement of irregular objects. Real data from industry is used to demonstrate the performance of the proposed methodology through various experiments, and the results are compared with those obtained by a genetic algorithm-based packing approach, and those generated from industrial practice. The results demonstrate the effectiveness of the proposed approach.