基于遗传-贪心混合搜索的人造板下料算法  

Cutting of wood-based panel based on genetic greedy hybrid search

作　　者：刘诚[1] 孙远升 花军[1] 姚嘉明 LIU Cheng;SUN Yuansheng;HUA Jun;YAO Jiaming(College of Mechanical and Electrical Engineering,Northeast Forestry University,Harbin 150040,China)

机构地区：[1]东北林业大学机电工程学院,哈尔滨150040

出　　处：《林业工程学报》2021年第4期127-133,共7页Journal of Forestry Engineering

基　　金：教育部高等学校博士学科点专项科研基金(博导类)(20130062110005)。

摘　　要：随着定制板式家具的需求扩大,对人造板加工下料中的智能排样技术提出了更高的要求,为此针对有"一刀切"约束的矩形板材排样问题展开研究。选择遗传算法与贪心算法混合使用,首先通过局部随机调序来改进有序种群初始方式,其次通过优化个体序列与工件初始序列中元素的位置映射关系,将其编码为基因序列,通过后检测排样的贪心策略扩大局部解搜索空间,最后以比例选择方式及最优保存策略改进优化遗传算子。结果显示:在非"一刀切"排样算例计算中,相比文献算法平均排样效果提高了2%,而计算时间缩减90%,相比文献同类型算法在获得相同求解效果时,迭代次数缩减50%;在有"一刀切"约束排样算例计算中,相比文献算法平均排样效果提高了2.7%,而平均余料数量减少了50%。试验证明:在进行非"一刀切"排样计算时,具有较高的时间效率和求解质量,同时具有较快的收敛速度,通过少量迭代便可获得较为满意的问题解;在进行有"一刀切"约束条件的排样计算同样可以获得较高的最优利用率,且在该类排样计算中可以明显降低余料碎化,使原料利用率得到进一步提升。With the expansion of the demand for customized panel furniture, the digital manufacturing has become the effective strategy to improve the competitiveness of furniture enterprises. So, the intelligent layout technology in wood-based panel cutting with high accuracy is put forward. This paper solved the layout problems of rectangular panels with "one size fits all" constraint. In this study, an intelligent algorithm was proposed to design cutting layouts for wood-based panels. Therefore, the utilization rate of raw materials and the efficiency of layout design were improved. The genetic algorithm and greedy algorithm were coupled to solve the optimal cutting problem. Firstly, the initial mode of the ordered population was improved by adding local random ordering. Next, the elements of individual sequence and initial sequence of the work piece had a location mapping relationship. The algorithm coded this relationship as gene sequence by optimizing the calculation procedure. The search space of local solution was expanded by the greedy strategy of post-detection nesting. In the end, the genetic operator was improved and optimized by the proportion selection method and optimal preservation strategy. The results showed that: 1) in the calculation of non "one size fits all" packing example, the average packing effect of this algorithm was improved by 2% compared with the other previous algorithms in literature. The calculation time of this algorithm was also reduced by 90%. When the propose method was compared with the same types of algorithms in literature, the iteration times of this algorithm were reduced by 50%. 2) when this method was compared with the algorithms in literature, the average nesting effect of this algorithm was improved by 2.7% and the average surplus convergence was reduced by 50% in the calculation of "one size fits all" packing example. The study concluded that: 1) when the proposed algorithm in this paper was carried out to calculate the non "one size fits all" packing example, it had advantage

关 键 词：遗传算法 贪心算法 人造板下料 一刀切排样 

分 类 号：S784[农业科学—林学]