求解矩形件下料问题的顺序启发式算法.pdfVIP

of this paper are as follows: Firstly, a quasi-human algorithm is introduced to generate cutting pattern which has the maximum value per unit area on basis of the optimal packing model. While using quasi-human algorithm to pack a rectangle blank, the blank always occupies a corner, and a hole as possible as it can. The quasi-human algorithm consists of greedy-algorithm and backtrack-algorithm. Greedy-algorithm always does the corner-occupying action with maximal caving degree. The backtrack-algorithm is used on the basis of greedy algorithm, once a blank is packed into the plate, the backtrack-algorithm is applied to confirm a corner-occupying action which has the maximal grade and then to do this action. On the basis of quasi-human algorithm, we improve the algorithm by reducing the number of corners and increasing the amount of strategy for selecting a corner-occupying action. Secondly, a value correction formula from the one-dimension packing problem is extended to solve the rectangle packing problem. The cutting patterns in the cutting solution are generated sequentially, when a pattern is generated by the quasi-human algorithm, every blanks in the plate of this pattern must be evaluating by the status of layout. The blank’s value is corrected by using the evaluation’s result combine with the prior information, and to get more reasonable by doing value correction repeatedly. Using traditional sequential heuristics procedure to generate cutting solution easily lead to local optimum situation, but combining with value correction can effectively avoid this situation and reach the global optimum. Several solutions are generated iteratively by implementing the sequential heuristics procedure based on value correction, and the solution with highest materi