公路交通科技
公路交通科技
공로교통과기
JOURNAL OF HIGHWAY AND TRANSPORTATION RESEARCH AND DEVELOPMENT
2010年
4期
149-154
,共6页
运输经济%动态选址-库存模型%遗传算法%克隆选择算法%粒子群优化算法%库存成本
運輸經濟%動態選阯-庫存模型%遺傳算法%剋隆選擇算法%粒子群優化算法%庫存成本
운수경제%동태선지-고존모형%유전산법%극륭선택산법%입자군우화산법%고존성본
transport economics%dynamic location-inventory model%genetic algorithm%clone selection algorithm%particle swarm optimization%inventory cost
为了解决传统配送中心选址没有同时考虑库存持有成本和决策环境的动态变化的问题,建立了一种新的模型.首先,利用两步骤近似方法获得(Q,R)库存策略下每一个周期配送中心的库存成本计算公式;然后,针对传统设施动态选址模型对选址成本的不恰当表示进行了修正,并与库存成本计算方法相结合,从而建立考虑库存成本的配送中心动态选址模型.最后,分别用遗传算法、克隆选择算法、粒子群优化算法求解所建立的模型,并从算法的精确度、稳定性、运算速度和收敛性比较了三种算法的性能.算例测试结果表明:所建立的模型是有效的;从总体上看,遗传算法的适应性要强于克隆选择算法和粒子群算法.
為瞭解決傳統配送中心選阯沒有同時攷慮庫存持有成本和決策環境的動態變化的問題,建立瞭一種新的模型.首先,利用兩步驟近似方法穫得(Q,R)庫存策略下每一箇週期配送中心的庫存成本計算公式;然後,針對傳統設施動態選阯模型對選阯成本的不恰噹錶示進行瞭脩正,併與庫存成本計算方法相結閤,從而建立攷慮庫存成本的配送中心動態選阯模型.最後,分彆用遺傳算法、剋隆選擇算法、粒子群優化算法求解所建立的模型,併從算法的精確度、穩定性、運算速度和收斂性比較瞭三種算法的性能.算例測試結果錶明:所建立的模型是有效的;從總體上看,遺傳算法的適應性要彊于剋隆選擇算法和粒子群算法.
위료해결전통배송중심선지몰유동시고필고존지유성본화결책배경적동태변화적문제,건립료일충신적모형.수선,이용량보취근사방법획득(Q,R)고존책략하매일개주기배송중심적고존성본계산공식;연후,침대전통설시동태선지모형대선지성본적불흡당표시진행료수정,병여고존성본계산방법상결합,종이건립고필고존성본적배송중심동태선지모형.최후,분별용유전산법、극륭선택산법、입자군우화산법구해소건립적모형,병종산법적정학도、은정성、운산속도화수렴성비교료삼충산법적성능.산례측시결과표명:소건립적모형시유효적;종총체상간,유전산법적괄응성요강우극륭선택산법화입자군산법.
In order to solve the problem that the inventory cost and the dynamic decision environment have not be considered simultaneously in traditional distribution center location models,a new model was developed. First,the formula of inventory cost in each period with (Q,R) rules was obtained through two steps approximately method. After that,the inaccurate formula of location cost in traditional dynamic facility location models was corrected,and a new dynamic distribution center location model with inventory cost was developed by integrating the formula of inventory cost into corrected dynamic facility location model. Finally,the model was solved by genetic algorithm,clone selection algorithm,particle swarm optimization respectively and the capacities of finding optimal solution,stability,counting speed and astringency were compared between these algorithms. The results of numerical example show that the model is effective and the genetic algorithm is most suitable in three algorithms for the problem.