A hybrid genetic algorithm for the two-dimensional single large object placement problem
Introduction
This paper considers the problem of cutting a plane rectangle into smaller rectangular items, each of a given size and value, with the objective of maximizing the total value of the items cut. This problem arises in many industrial applications; for example, the cutting of steel, glass, wood or paper and the placement of adverts in newspapers and magazines. The related problem of minimizing the amount of wasted material produced by cuts can be converted into the value maximization problem by simply taking the value of each item to be proportional to its area.
The two-dimensional cutting problem can also be viewed as the problem of packing smaller items into a larger stock rectangle. Packing problems are common in the optimization of stock areas in industry layouts. In this paper, the terms cutting and packing will be used interchangeably.
The above problem studied in this paper is known in the literature as the two-dimensional single large object placement problem (2SLOPP). The name of the problem is derived from the typology developed by Wäscher et al. [52]. Note that, in the same typology, the related problem in which there is a strongly heterogeneous assortment of items is defined as the two-dimensional knapsack problem (2KP).
We consider the constrained two-dimensional non-guillotine cutting problem in which each item must be cut with its edges always parallel to the edges of the master surface (i.e., orthogonal cuts) and it must have a fixed orientation (i.e., no rotation is allowed). Cutting problems in which the orientation of the items is fixed arise for materials that are not isomorphic, as wood or corrugated sheet. The optimal cutting patterns are not restricted to be of the guillotine type (a guillotine cut on a rectangle runs from one edge of the rectangle to the opposite edge, parallel to the two remaining edges). Moreover, the constrained form of this problem imposes restrictions on the maximum number of items of each type (i.e., same size and value) required to be cut.
In this paper, we present a simple but very effective heuristic approach for solving the 2SLOPP. In a preliminary phase, simple upper bounds are computed and initial solutions are obtained through greedy algorithms. Then a genetic search, which uses parent selections, elitist theory, immigration and different crossover operators, is performed. The genetic approach is hybridized with an on-line heuristic. The computational results reported show that the algorithm finds the best solution in a fast computational time for most of the cases and, hence, outperforms the other metaheuristics in the literature.
The rest of the paper is organized as follows: In Section 2, we survey the main results in the literature. In Section 3, we briefly describe the problem and its main characteristics. In Section 4, we present a simple upper bound from the literature and propose new greedy algorithms for the computation of lower bounds. Section 5 presents our hybrid genetic approach and Section 6 reports extensive computational results of this approach for a wide range of problem instances.
Section snippets
Literature review
Many surveys are available for multidimensional cutting and packing problems. We refer to Dyckoff [17], Haessler and Sweeney [28], Sweeney and Paternoster [51], Dowsland and Dowsland [16] and Lodi et al. [41].
After the seminal work of Gilmore and Gomory [24], a number of authors have addressed the two-dimensional cutting problem in its different forms. The unconstrained non-guillotine problem has been considered by a few authors, namely, Arenales and Morabito [2] who proposed an approach based
Problem description
The 2SLOPP can be defined as follows: A large rectangular master surface, here defined as S, of width W0 and length L0 has to be cut into a number of smaller rectangular items chosen from a set of m available types. Each item type j is defined by width wj, length lj and value vj, for j = 1, … , m. The objective is to construct a cutting pattern for S with the highest possible total value.
The constrained 2SLOPP is considered, in which the number of replicates of each item produced should not exceed a
A one-dimensional knapsack relaxation
Consider a natural relaxation of the 2SLOPP, given by the one-dimensional knapsack (1KP), in which each item has an associated value vj and a weight dj equal to the area of the item j in the 2SLOPP (dj = wjlj) for j = 1, … , M, and the knapsack capacity D is equal to the area of S in the 2SLOPP (D = W0L0). The 1KP can be formally stated as
For a 2SLOPP instance I, let z(I) denote the optimal solution value of I and zUB the upper bound on z(I) corresponding to
Genetic algorithm
In this section we will outline a hybrid genetic algorithm, referred to as GA2SLOPP, which we have developed for the 2SLOPP. The following elements of GA2SLOPP will be discussed:
- •
Representation: How a solution to the 2SLOPP is represented;
- •
Fitness: Assessing the value of a solution;
- •
Evolution process: Elitist criteria, reproduction and immigration;
- •
Crossover: Having a child from parents;
- •
On-line heuristic: Finding a solution.
Computational results
The heuristic algorithm GA2SLOPP was coded in FORTRAN 90 and run on a Pentium IV 1700 MHz (Windows 2000 operating system), using a set of both small and large-size test problems and a particular problem of packing squares.
Conclusions
In this paper we have presented a new hybrid genetic algorithm for the two-dimensional knapsack problem. The algorithm was evaluated on a very large number of test problems taken from the literature of varying size and complexity. The reported computational results indicate that as problem size increases, our heuristic performs better, getting results that are (empirically) very close to optimality in small computational times. Indeed, for the largest problems solved, our results outperform
Acknowledgements
We thank the Ministero dell’Istruzione, dell’Università e della Ricerca (MIUR) and the Consiglio Nazionale delle Ricerche (CNR), Italy, for the support given to this project. The computational experiments have been executed at the Laboratory of Operations Research (LabOR) of the University of Bologna.
References (53)
- et al.
An and/or-graph approach to the solution of two-dimensional guillotine cutting problems
European Journal of Operational Research
(1995) A population heuristic for constrained two-dimensional non-guillotine cutting
European Journal of Operational Research
(2004)- et al.
On the 2-dimensional knapsack problem
Operations Research Letters
(2004) - et al.
An exact algorithm for the orthogonal, 2-d cutting problems using guillotine cuts
European Journal of Operational Research
(1995) - et al.
Packing problems
European Journal of Operational Research
(1992) A typology of cutting and packing problems
European Journal of Operational Research
(1990)- et al.
An exact algorithm for general, orthogonal, two-dimensional knapsack problems
European Journal of Operational Research
(1995) - et al.
Cutting stock problems and solution procedures
European Journal of Operational Research
(1991) - et al.
An optimal algorithm for rectangle placement
Operations Research Letters
(1999) - et al.
An empirical investigation of meta-heuristic and heuristic algorithms for a 2d packing problem
European Journal of Operational Research
(2001)
On genetic algorithms for the packing of polygons
European Journal of Operational Research
Developing a simulated annealing algorithm for the cutting stock problem
Computers and Industrial Engineering
Applications of genetic search and simulated annealing to the two-dimensional non-guillotine cutting stock problem
Computers and Industrial Engineering
An improved BL-algorithm for genetic algorithm of the orthogonal packing of rectangles
European Journal of Operational Research
Two-dimensional packing problems: A survey
European Journal of Operational Research
Approximation algorithms for the oriented two-dimensional bin packing problem
European Journal of Operational Research
Genetic algorithm crossover operators for ordering applications
Computers and Operations Research
Algorithms for unconstrained two-dimensional guillotine cutting
Journal of the Operational Research Society
An exact two-dimensional non-guillotine cutting tree search procedure
Operations Research
Or-library: Distributing test problems by electronic mail
Journal of the Operational Research Society
New upper bounds for the two-dimensional orthogonal non guillotine cutting stock problem
IMA Journal of Management Mathematics
A new placement heuristic for the orthogonal stock-cutting problem
Operations Research
An algorithm for two-dimensional cutting problems
Operations Research
A Handbook of Genetic Algorithms
Cited by (36)
Knapsack problems — An overview of recent advances. Part II: Multiple, multidimensional, and quadratic knapsack problems
2022, Computers and Operations ResearchCitation Excerpt :Beasley (2004) proposed a successful population heuristic based upon a non-linear formulation and tested it on several benchmarks involving up to 4000 items. Improved computational results were obtained by Alvarez-Valdés et al. (2007) (Tabu search), Hadjiconstantinou and Iori (2007) (genetic algorithm), Leung et al. (2012) (simulated annealing), Kierkosz and Luczak (2014) (hybrid evolutionary algorithm), and Shiangjen et al. (2018) (iterative bidirectional heuristic). Exact solution.
Exact solution techniques for two-dimensional cutting and packing
2021, European Journal of Operational Research3D overlapped grouping Ga for optimum 2D guillotine cutting stock problem
2014, Alexandria Engineering JournalCitation Excerpt :The cutting stock problem (CSP) is one of the oldest and most studied problems in the field of combinatorial optimization. Cutting stock problems in all their variants have been extensively treated in literature in the last decades [12,13,25]. Due to their complexity, most approaches found in literature relied on heuristics.
Genotype-phenotype heuristic approaches for a cutting stock problem with circular patterns
2013, Engineering Applications of Artificial IntelligenceCitation Excerpt :Burke et al. (2009) approach the orthogonal cutting stock problem with simulated annealing, and they obtain good solutions in a few minutes of execution time for several small instances taken from the literature. In addition, using a genetic algorithm, Hadjiconstantinou and Iori (2007) solve instances of large-size rectangular cutting problems in a matter of minutes. Section 2 presents the mathematical model for the SSP.
Two-dimensional strip packing problem with load balancing, load bearing and multi-drop constraints
2013, International Journal of Production EconomicsCitation Excerpt :The first one consists of an adapted version of instances cgcut01-cgcut03 and ngcut01-ngcut12 which are available at OR-Library (Beasley, 1990). These instances have been used, for example, by Caprara and Monaci (2004) and Hadjiconstantinou and Iori (2007) when solving the Two-dimensional Constrained Knapsack problem for guillotine and non-guillotine cuts, respectively. In these instances we consider only one copy of each item, that is, if item i has demand of three pieces in the original instance, now it has demand of only one piece.