CN103399868A - 一种外贸物流路径的优化方法 - Google Patents

一种外贸物流路径的优化方法 Download PDF

Info

Publication number
CN103399868A
CN103399868A CN2013102849668A CN201310284966A CN103399868A CN 103399868 A CN103399868 A CN 103399868A CN 2013102849668 A CN2013102849668 A CN 2013102849668A CN 201310284966 A CN201310284966 A CN 201310284966A CN 103399868 A CN103399868 A CN 103399868A
Authority
CN
China
Prior art keywords
upsi
psi
algorithm
partiald
node
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN2013102849668A
Other languages
English (en)
Other versions
CN103399868B (zh
Inventor
初佃辉
叶允明
李春山
周学权
王德泉
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Harbin Institute of Technology
Harbin Institute of Technology Weihai
Original Assignee
Harbin Institute of Technology Weihai
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Harbin Institute of Technology Weihai filed Critical Harbin Institute of Technology Weihai
Priority to CN201310284966.8A priority Critical patent/CN103399868B/zh
Publication of CN103399868A publication Critical patent/CN103399868A/zh
Application granted granted Critical
Publication of CN103399868B publication Critical patent/CN103399868B/zh
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本发明涉及一种外贸物流路径的优化方法,其解决了目前计算方法事实操作可行性不强,面对大规模的物流网络无法精确且成本高,无自适用学习能力。其通过建立无向图模型推到出期望值模型进行计算得出最优路径。其可广泛应用物流运输领域。

Description

一种外贸物流路径的优化方法
技术领域
本发明涉及一种路径确定方法,具体说是一种外贸物流路径的优化方法。
背景技术
随着经济全球化迅速发展,物流业已经成为商业环节的一个重要组成部分。在各种不同需求的推动下,一些功能组件应运而生的并在现代物流业中发挥重要作用,如电子跟踪、仓储、资源分配。同时,随着服务中心、信息处理中心、资源配置中心的出现,整个供应链需要建立无缝连接,以使得物流网络能够高效运转。因此,由于功能组件引入,路径最优化的复杂性日益增强且相应生成的路径组合变得更加庞大。
物流是指为了满足客户的需求,以最低的成本,通过运输、保管、配送等方式,实现原材料、半成品、成品或相关信息进行由商品的产地到商品的消费地的计划、实施和管理的全过程。而外贸物流是指为一个国家(地区)与另一个国家(地区)为满足客户的各种需要,通过某种运输方式,实现两个国家(地区)相关产品交易。全部物流活动是在线路和节点进行的,物流网络指由执行物流活动使命的线路和执行物流停顿使命的节点两种基本元素所组成的网络。外贸物流网络包括六种类型的实体(节点),如起点(出口商),代理,运输公司,仓储站,运输公司和目的地(进口商)。运输是物流中最重要的功能要素之一,物流合理化在很大程度上依赖于运输合理化。运输路径安排问题(LRP)是集成化物流系统中路径优化问题的一个重要分支,是任何现代物流系统必须要面临的问题。
在物流系统中路径优化问题中,最重要的问题是在所有可能的路径中寻找最佳的运输路径以使其运输成本最小。现有的方法是在整个运输网络中进行组合优化以寻求帮助,本质上,这是个NP-Hard问题,因此,这些方法只能得到近似解。随着物流网络规模的迅速增长,例如在外贸行业中,在不同的法律约束或面对用户的不同要求,通过此方法获得高效率和有效的解决方案变得越来越困难。现有方法中还有一些启发式算法或基于智能代理的算法,如遗传算法、蚁群算法和免疫算法。在起步阶段,针对这个问题,在许多研究工作中广泛采用遗传算法(GA)以优化物流路线,GA是一种基于自适应启发式搜索算法以进化中的自然选择和遗传变异前提。近几年,进化算法(EA),试图利用类似的技术,如遗传,变异,选择和交叉解决路径优化问题。免疫算法是模仿免疫系统,以解决多模态函数优化问题的遗传算法的变种。然后有研究者提出了基于智能代理技术的蚁群优化算法(ACO)用于解决组合优化问题,它模拟蚂蚁的觅食行为时产生感知信息素以帮助其他蚂蚁成功找到食物。在ACO算法中,首先要构建一定数目的虚拟蚂蚁,在完全连接图中按照某种规则出发,各自独立地根据信息素和启发式信息,采用一个概率规则选择下一步的移动,直到建立优化问题一个完整的解。
虽然上述算法已广泛应用于各种领域,如图形着色,路由选择和旅行商问题,但仍存在以下缺陷。第一,这些方法的问题在于假定物流网络每条边的成本是已知的,事实操作中,这是不可行的,因为这样大规模的物流网络无法精确的计算每条边的成本。第二,面对大量的约束,直接对整个物流网络进行优化是非常困难的。第三,由于缺乏自适应学习能力,无法从历史数据中获取知识。
发明内容
本发明就是为了解决上述技术问题,提供一种基于图模型的外贸物流路径的优化方法。
本发明的技术方案是,提供一种物流运输路径确定方法,包括一下步骤:
(1)识别外贸物流网络中的实体;
(2)构建外贸物流网络;
(3)将所述外贸物流网络抽象为层次无向图;
(4)针对所述层次无向图,面对不同场景,计算最优路径。
优选地,其特征在于所述实体包括:起点,代理,运输公司,仓储站,运输公司和目的地。
优选地,层次无向图中,不同层次的节点抽象为无向图的节点集合V,不同层次节点之间的连线抽象为无线图边的集合E,形成G=<V,E>;节点点集V可以分为k个不相交的子集:V=S1∪S2∪…∪Sk,用vi来表示一个节点,则有
Figure BDA00003477398900034
,边集E={<Vi,Vi+1|Vi∈Si,Vi+1∈Si+1>},i∈{1,...,k-1},每条边e=<vi,vj>的权为ωij表示该边在物流路径上的概率。
优选地,步骤(4)中的计算方法如下:
A:基于随机游走模型的运输路线算法
所述层次无向图中,每对<Si,Si+1>生成一个转移概率矩阵M;因此,存在5个转移概率矩阵:M1,M2,M3,M4和M5,矩阵Mi为Si上的转移概率矩阵;定义如下的优化路径迭代方程:
&upsi; 1 = ( 1 - c ) M 1 &upsi; 2 + cp &upsi; 2 = 1 2 M 1 T &upsi; 1 + 1 2 M 2 &upsi; 3 &upsi; 3 = 1 2 M 2 T &upsi; 2 + 1 2 M 3 &upsi; 4 &upsi; 4 = 1 2 M 3 T &upsi; 3 + 1 2 M 4 &upsi; 5 &upsi; 5 = 1 2 M 5 T &upsi; 3 + 1 2 M 4 &upsi; 5 &upsi; 6 = ( 1 - c ) M 5 &upsi; 2 + cq - - - ( 1 )
向量p和q具有初始值,c是一个常数,根据经验设定为0.5;
转移概率定义为:
f &psi; ( &upsi; i , &upsi; j ) = 1 1 + exp ( - F ( &upsi; i , &upsi; j , &psi; ) ) = 1 1 + exp ( - &Sigma; k = 1 n &psi; k ( &upsi; i k - &upsi; j k ) 2 ) . - - - ( 2 )
参数ψ可以采用最大似然估计,对数似然函数是:
Figure BDA00003477398900033
,其中m是边的数量;最大化步骤可表示为:
&PartialD; F ( &upsi; i , &upsi; j , &psi; ) &PartialD; &psi; k = sim ( &upsi; i k , &upsi; j k ) &PartialD; L &PartialD; &psi; k = ( - 1 ) &Sigma; m exp ( - F ( &upsi; i , &upsi; j , &psi; ) ) 1 + exp ( - F ( &upsi; i , &upsi; j , &psi; ) ) &PartialD; F ( &upsi; i , &upsi; j , &psi; ) &PartialD; &psi; k = ( - 1 ) &Sigma; m exp ( - F ( &upsi; i , &upsi; j , &psi; ) ) 1 + exp ( - F ( &upsi; i , &upsi; j , &psi; ) ) sim ( &upsi; i k , &upsi; j k ) &psi; t = &psi; t - 1 + &eta; &PartialD; L &PartialD; &psi; , - - - ( 3 )
其中η是迭代参数,当|ψtt-1|小于预定义的ε值时,迭代结束;当没有特殊要求或约束时,可以通过RWTR算法获得的最佳路径;该算法直接采用随机游走模型优化路径,算法描述如下:
RWTR算法:
Figure BDA00003477398900042
或者B:面向约束的运输路径算法
CTR算法:
Figure BDA00003477398900043
Figure BDA00003477398900051
或者C:增量式运输路径算法
一旦接收到增量数据集Dt,转换概率矩阵M1,M2,M3,M4,M5马上被公式(4)中更新;
M i &prime; = | D | | D | + | D t | M i + | D t | | D | + | D t | M it , - - - ( 4 )
其中,Mit是由Dt计算出来的转换概率矩阵;
ICTR算法:
Figure BDA00003477398900053
本发明的有益效果是,实际操作可行性更强,适用于大规模物流网络;针对大量约束条件,计算准确,效率高;无法能够从历史数据中获取知识,拥有自适应学习能力。
本发明进一步的特征和方面,将在以下参考附图的具体实施方式的描述中,得以清楚地记载。
附图说明
图1为外贸物流网络图;
图2为无向图模型图;
图3为本发明的流程图。
图中符号说明:
1.起点;2.代理;3.运输公司;4.仓储站;5.运输公司;6.目的地。
具体实施方式
以下参照附图,以具体实施例对本发明作进一步详细说明。
参照图1、2和3,本发明运用图模型进行算法求解最优路径。
第一步,识别外贸物流网络中的实体,构建外贸物流网络,建立模型。
外贸物流网络包括六种类型的实体(节点),如起点1、代理2、运输公司3、仓储站4、运输公司5和目的地6,识别这六种实体,即六个节点,寻找出口商(第一个节点),通过出口代理雇用某个满足出口的要求船公司,然后一些场站被选择作为商品的暂时存储场所,运输公司负责将商品从场站运送到目的地(最后一个节点)。起点1可以是出口商,目的地6可以是进口商。
这样看来,外贸物流网络是一个层次结构,每个节点属于一个唯一的类型,同一类型的节点构成同一层;每一层在物流链中有唯一的位置;一个完整的物流路径包括依次连接相邻层的边。
根据上述网络建立层次无向图模型的方法如下:
节点集合V,不同层次节点之间的连线边的集合E,形成G=(V,E)。节点集合V可以分为k个不相交的子集:V=S1∪S2∪…∪Sk,用vi来表示一个节点,则有
Figure BDA00003477398900061
,边的集和E={<Vi,Vi+1|Vi∈Si,Vi+1∈Si+1>},i∈{1,...,k-1},每条边e=(vi,vj)的权为ωij,表示该边在物流路径上的概率。
起点1为20个,目的地6为20个,代理2、运输公司3、仓储站4、运输公司5各为200个。因为在实际的物流应用中起点1是不确定的,比如说相同的起始点、目的地,而其他属性(重量,要求时间)不同。而在图算法中,各个节点是必须是确定的。所以每个起点1将被扩展成为4个确定的点,这4个点可以包含一个起点1的所有情况(在实际应用中可以扩展成为更多更具体的初始点)。总共有900个节点,80个起点1,20个目的地6,其他内部子集,每个子集中包含200个节点,如表1所示。
表1
节点集 编号
起点(出口商)1 1-80
代理2 101-300
运输公司3 301-500
仓储站4 501-700
运输公司5 701-900
目的地(进口商)6 81-100
如表2所示,我们使用影响物流选择的主要属性来描述节点:每个节点包含5个属性(起始点,目的地,最大负载量,最快处理时间,花费价格)。
表2
节点集 例子
起点(出口商)1 (1,1,28,11,-1)
代理2 (-1,-1,88,22,74)
运输公司3 (-1,5,41,18,24)
仓储站4 (-1,-1,47,21,20)
运输公司5 (12,12,43,14,33)
目的地(进口商)6 (-1,5,-1,-1,-1)
如表2所示,给出了各个节点子集的一个例子。在起点(出口商)1中(1,1,28,11,-1)节点表示,货物的起始地点是1,目的地点是1,重量是28,要求11天之内送达,-1表示可以为任意值。代理2子集中的节点(-1,-1,88,22,74)表示,此出口代理最大能处理重量为88的货物,最快处理时间是22,价格是74,这个代理可以处理从任意起始点到任意目的地的货物。
上述物流网络图中相邻子集间的节点是全向量,各个子集内部没有变。
第二步,根据不同情况使用算法计算最优路径,主要分为三种情况。
1)第一种情况是没有约束条件的情况下基于随机游走模型的运输路线算法(RWTR):
无向图模型中每对<Si,Si+1>生成一个转移概率矩阵M。因此,存在5个转移概率矩阵:M1,M2,M3,M4,M5,矩阵Mi为Si上的转移概率矩阵。定义如下的优化路径迭代方程:
&upsi; 1 = ( 1 - c ) M 1 &upsi; 2 + cp &upsi; 2 = 1 2 M 1 T &upsi; 1 + 1 2 M 2 &upsi; 3 &upsi; 3 = 1 2 M 2 T &upsi; 2 + 1 2 M 3 &upsi; 4 &upsi; 4 = 1 2 M 3 T &upsi; 3 + 1 2 M 4 &upsi; 5 &upsi; 5 = 1 2 M 5 T &upsi; 3 + 1 2 M 4 &upsi; 5 &upsi; 6 = ( 1 - c ) M 5 &upsi; 2 + cq - - - ( 1 )
向量p和q具有初始值,c是一个常数,根据经验设定为0.5。
转移概率定义为:
f &psi; ( &upsi; i , &upsi; j ) = 1 1 + exp ( - F ( &upsi; i , &upsi; j , &psi; ) ) = 1 1 + exp ( - &Sigma; k = 1 n &psi; k ( &upsi; i k - &upsi; j k ) 2 ) . - - - ( 2 )
参数ψ可以采用最大似然估计,对数似然函数是
l ( &psi; ) = log ( &Pi; m f &psi; ( &upsi; i , &upsi; j ) )
其中m是边的数量,最大化步骤可表示为
&PartialD; F ( &upsi; i , &upsi; j , &psi; ) &PartialD; &psi; k = sim ( &upsi; i k , &upsi; j k ) &PartialD; L &PartialD; &psi; k = ( - 1 ) &Sigma; m exp ( - F ( &upsi; i , &upsi; j , &psi; ) ) 1 + exp ( - F ( &upsi; i , &upsi; j , &psi; ) ) &PartialD; F ( &upsi; i , &upsi; j , &psi; ) &PartialD; &psi; k = ( - 1 ) &Sigma; m exp ( - F ( &upsi; i , &upsi; j , &psi; ) ) 1 + exp ( - F ( &upsi; i , &upsi; j , &psi; ) ) sim ( &upsi; i k , &upsi; j k ) &psi; t = &psi; t - 1 + &eta; &PartialD; L &PartialD; &psi; , - - - ( 3 )
其中η是迭代参数,当|ψtt-1|小于预定义的ε值时,迭代结束。
此算法只考虑出口货物的起始点和目的地以及运输价格,其他的不考虑。根据这个要求,实验人工选择产生了100条历史数据,下面列出其中的5条:
71,171,431,591,751,15
79,174,434,594,754,16
49,219,379,619,879,5
60,120,500,595,755,7
63,173,433,593,753,14
其中,71,171,431,591,751,15表示货物从节点71代表的起始点出发,经过出口代理171,船公司431,场站591,运输公司753最后达到目的地15。
接着,我们根据论文中的公式(3)算出参数ψ=[0.005;0.5608;0.0732;0.0683;0.2477]。然后根据公式(2)产生图中各个子集之间的状态转移矩阵。矩阵M12的大小是80*200,矩阵M56的大小是80*200,其他矩阵的大小是200*200。
下面展示了M12的一行,此行中的每个元素都表示节点被选中成为一条路径上的边的概率。
0.0048812,0.005031,0.005031,0.005031,0.005031,0.0049189,0.0050687,0.0050687,0.0050687,0.0050687,0.0048329,0.0049826,0.0049826,0.0049826,0.0049826,0.00488,0.0050298,0.0050298,0.0050298,0.0050298,0.0048673,0.0050171,0.0050171,0.0050171,0.0050171,0.0049076,0.0050574,0.0050574,0.0050574,0.0050574,0.0048353,0.004985,0.004985,0.004985,0.004985,0.0048109,0.0049607,0.0049607,0.0049607,0.0049607,0.0048979,0.0050477,0.0050477,0.0050477,0.0050477,0.0048992,0.005049,0.005049,0.005049,0.005049,0.0048409,0.0049907,0.0049907,0.0049907,0.0049907,0.0048274,0.0049772,0.0049772,0.0049772,0.0049772,0.0048884,0.0050382,0.0050382,0.0050382,0.0050382,0.0048829,0.0050327,0.0050327,0.0050327,0.0050327,0.0048598,0.0050096,0.0050096,0.0050096,0.0050096,0.0048049,0.0049546,0.0049546,0.0049546,0.0049546,0.0048999,0.0050498,0.0050498,0.0050498,0.0050498,0.0049389,0.0050887,0.0050887,0.0050887,0.0050887,0.0048704,0.0050202,0.0050202,0.0050202,0.0050202,0.0048604,0.0050102,0.0050102,0.0050102,0.0050102,0.0049345,0.0050843,0.0050843,0.0050843,0.0050843,0.0049078,0.0050576,0.0050576,0.0050576,0.0050576,0.0048634,0.0050132,0.0050132,0.0050132,0.0050132,0.0049282,0.005078,0.005078,0.005078,0.005078,0.0048997,0.0050495,0.0050495,0.0050495,0.0050495,0.0049662,0.005116,0.005116,0.005116,0.005116,0.0047989,0.0049487,0.0049487,0.0049487,0.0049487,0.0048459,0.0049956,0.0049956,0.0049956,0.0049956,0.0048802,0.00503,0.00503,0.00503,0.00503,0.0049896,0.0051394,0.0051394,0.0051394,0.0051394,0.0048357,0.0049855,0.0049855,0.0049855,0.0049855,0.004896,0.0050458,0.0050458,0.0050458,0.0050458,0.0049191,0.005069,0.005069,0.005069,0.005069,0.0049233,0.0050731,0.0050731,0.0050731,0.0050731,0.004824,0.0049737,0.0049737,0.0049737,0.0049737,0.0048283,0.0049781,0.0049781,0.0049781,0.0049781,0.0049089,0.0050587,0.0050587,0.0050587,0.0050587,0.0049189,0.0050687,0.0050687,0.0050687,0.0050687,0.0048492,0.004999,0.004999,0.004999,0.004999,0.0048837,0.0050335,0.0050335,0.0050335,0.0050335。
产生状态转移矩阵后,RWTR算法就可以运行起来了,给定了初始值,比如说我们在起始点2,想发送一批货物到目的地3,货物的重量是70.我们匹配后发现,此情况对应物流图的节点38,于是我们有了起始点确定了起始点38,目的节点83.由此我们在每个子集上构建一个初始向量起始向量为v1=(0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)。第38为1,表示起始点。目的地向量为v6=(0,0,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0)。第三位为1,表示目的地3。V2=V3=V4=V5=(0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005,0.005)。其中向量每个元素值代表了子集中的一个节点,初始值都是1/200。
RWTR算法结束后我们得到了V2,V3,V4,V5。我们取每个向量中的最大值元素对应的节点,认为这个节点是物流路径中的一个节点。综合起始点和目的地点,可以得到一条优化的物流路径:
(38,197,397,652,712,83)。
算法步骤如表3所示:
表3:
2)有约束情况下的运输路径算法(CTR):
CTR算法,该算法中状态转移矩阵和V向量的初始值设定都和RWTR相似。不同点在于我们产生优化物流路径的做法。当算法结束后我们得到了V2,V3,V4,V5。因为这个算法中考虑了时间等限制条件,我们在选取V2中的物流节点时,不是直接选取元素最大值对应的节点。而是对V2中节点根据元素值进行排序。然后从高到低选择第一个满足条件的节点。
当存在约束条件时,如运输时间和价格等,RWTR可以扩展为CTR算法。在该算法中全局约束可以分割成几个基于对历史数据的统计估计片段,这些片段的数量被设置为在网络中层的数量,每个约束片段对应的一层。根据历史统计数据初始化所有层,用同样的方法处理的其余部分约束。
算法步骤如表4所示:
表4:
Figure BDA00003477398900131
3)考虑传入数据的影响时运用增量式运输路径算法(ICTR):
算法CTR相比,关键的不同在于我们有了新的增量数据。我们在新数据下产生了新的状态转移矩阵,然后根据公式(4),生成增量状态转移矩阵。其他的优化路径的产生与CTR相同。
一旦接收到增量数据集Dt,转换概率矩阵M1,M2,M3,M4,M5马上被公式(4)中更新。
M i &prime; = | D | | D | + | D t | M i + | D t | | D | + | D t | M it , - - - ( 4 )
其中Mit是由Dt计算出来的转换概率矩阵。
算法步骤如表5所示:
表5:
Figure BDA00003477398900141

Claims (4)

1.一种外贸物流路径的优化方法,其特征在于包括以下步骤:
(1)识别外贸物流网络中的实体;
(2)构建外贸物流网络;
(3)将所述外贸物流网络抽象为层次无向图;
(4)针对所述层次无向图,面对不同场景,计算最优路径。
2.根据权利要求1所述的外贸物流路径的优化方法,其特征在于所述实体包括:起点,代理,运输公司,仓储站,运输公司和目的地。
3.根据权利要求1所述的外贸物流路径的优化方法,其特征在于所述层次无向图中,不同层次的节点抽象为无向图的节点集合V,不同层次节点之间的连线抽象为无线图边的集合E,形成G=<V,E>;节点点集V可以分为k个不相交的子集:V=S1∪S2∪…∪Sk,用Vi来表示一个节点,则有
Figure FDA00003477398800012
,边集E={<Vi,Vi+1|Vi∈Si,Vi+1∈Si+1>},i∈{1,...,k-1},每条边e=<vi,vj>的权为ωij表示该边在物流路径上的概率。
4.根据权利要求3所述的外贸物流路径的优化方法,其特征在于所述步骤(4)中的计算方法如下:
A:基于随机游走模型的运输路线算法
所述层次无向图中,每对<Si,Si+1>生成一个转移概率矩阵M;因此,存在5个转移概率矩阵:M1,M2,M3,M4和M5,矩阵Mi为Si上的转移概率矩阵;定义如下的优化路径迭代方程:
&upsi; 1 = ( 1 - c ) M 1 &upsi; 2 + cp &upsi; 2 = 1 2 M 1 T &upsi; 1 + 1 2 M 2 &upsi; 3 &upsi; 3 = 1 2 M 2 T &upsi; 2 + 1 2 M 3 &upsi; 4 &upsi; 4 = 1 2 M 3 T &upsi; 3 + 1 2 M 4 &upsi; 5 &upsi; 5 = 1 2 M 5 T &upsi; 3 + 1 2 M 4 &upsi; 5 &upsi; 6 = ( 1 - c ) M 5 &upsi; 2 + cq - - - ( 1 )
向量p和q具有初始值,c是一个常数,根据经验设定为0.5;
转移概率定义为:
f &psi; ( &upsi; i , &upsi; j ) = 1 1 + exp ( - F ( &upsi; i , &upsi; j , &psi; ) ) = 1 1 + exp ( - &Sigma; k = 1 n &psi; k ( &upsi; i k - &upsi; j k ) 2 ) . - - - ( 2 )
参数ψ可以采用最大似然估计,对数似然函数是:
Figure FDA00003477398800022
,其中m是边的数量;最大化步骤可表示为:
&PartialD; F ( &upsi; i , &upsi; j , &psi; ) &PartialD; &psi; k = sim ( &upsi; i k , &upsi; j k ) &PartialD; L &PartialD; &psi; k = ( - 1 ) &Sigma; m exp ( - F ( &upsi; i , &upsi; j , &psi; ) ) 1 + exp ( - F ( &upsi; i , &upsi; j , &psi; ) ) &PartialD; F ( &upsi; i , &upsi; j , &psi; ) &PartialD; &psi; k = ( - 1 ) &Sigma; m exp ( - F ( &upsi; i , &upsi; j , &psi; ) ) 1 + exp ( - F ( &upsi; i , &upsi; j , &psi; ) ) sim ( &upsi; i k , &upsi; j k ) &psi; t = &psi; t - 1 + &eta; &PartialD; L &PartialD; &psi; , - - - ( 3 )
其中η是迭代参数,当|ψtt-1|小于预定义的ε值时,迭代结束;当没有特殊要求或约束时,可以通过RWTR算法获得的最佳路径;该算法直接采用随机游走模型优化路径,算法如下表所示:
Figure FDA00003477398800024
或者B:面向约束的运输路径算法,如下表所示的CTR算法
或者C:增量式运输路径算法
一旦接收到增量数据集Dt,转换概率矩阵M1,M2,M3,M4,M5马上被公式(4)中更新;
M i &prime; = | D | | D | + | D t | M i + | D t | | D | + | D t | M it , - - - ( 4 )
其中,Mit是由Dt计算出来的转换概率矩阵;
如下表所示的ICTR算法:
Figure FDA00003477398800033
CN201310284966.8A 2013-07-08 2013-07-08 一种外贸物流路径的优化方法 Active CN103399868B (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201310284966.8A CN103399868B (zh) 2013-07-08 2013-07-08 一种外贸物流路径的优化方法

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201310284966.8A CN103399868B (zh) 2013-07-08 2013-07-08 一种外贸物流路径的优化方法

Publications (2)

Publication Number Publication Date
CN103399868A true CN103399868A (zh) 2013-11-20
CN103399868B CN103399868B (zh) 2016-08-10

Family

ID=49563498

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201310284966.8A Active CN103399868B (zh) 2013-07-08 2013-07-08 一种外贸物流路径的优化方法

Country Status (1)

Country Link
CN (1) CN103399868B (zh)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105205637A (zh) * 2015-09-24 2015-12-30 天津维金斯环保科技有限公司 一种物流管理系统及管理方法
CN105589872A (zh) * 2014-10-23 2016-05-18 阿里巴巴集团控股有限公司 图搜索结果获取方法及装置
CN106339831A (zh) * 2015-07-09 2017-01-18 阿里巴巴集团控股有限公司 用于为业务获取有效路径的方法及装置
CN106919998A (zh) * 2015-12-28 2017-07-04 航天信息股份有限公司 一种基于aoe网络的物流模型及方法
CN108022060A (zh) * 2016-10-31 2018-05-11 菜鸟智能物流控股有限公司 物流网络的负载参数优化方法及装置
CN109408921A (zh) * 2018-10-12 2019-03-01 西南交通大学 基于物流路径网络的复杂离散制造系统物流建模仿真方法
CN111445093A (zh) * 2020-04-23 2020-07-24 长春工程学院 一种输电线路应急抢修路径优化系统及方法
CN112133443A (zh) * 2020-08-25 2020-12-25 上海大学 一种基于随机游走的时序网络免疫方法
WO2022018552A1 (en) * 2020-07-24 2022-01-27 International Business Machines Corporation Automated generation of optimization model for system-wide plant optimization
CN115660244A (zh) * 2022-12-27 2023-01-31 北京京东振世信息技术有限公司 路线信息生成方法、装置、设备和介质
CN116109023A (zh) * 2023-02-07 2023-05-12 广东工业大学 一种针对邮差派件的交通路径优化方法、装置及存储介质

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6477515B1 (en) * 1999-08-11 2002-11-05 The United States Of America As Represented By The Secretary Of The Navy Efficient computation of least cost paths with hard constraints
US20060015830A1 (en) * 2004-07-14 2006-01-19 Winfried Schwarzmann Systems, methods, and articles of manufacture for flexible path optimization
CN1741052A (zh) * 2005-09-22 2006-03-01 上海交通大学 车辆配载及路径优化系统
CN101055629A (zh) * 2006-04-11 2007-10-17 上海宝钢冶金建设公司 适于满载配送的车辆调度方法
CN101673382A (zh) * 2009-10-21 2010-03-17 北京交通大学 一种农资连锁经营物流配送配载组合优化方法
CN202383745U (zh) * 2011-07-13 2012-08-15 惠州天缘电子有限公司 物流信息服务系统

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6477515B1 (en) * 1999-08-11 2002-11-05 The United States Of America As Represented By The Secretary Of The Navy Efficient computation of least cost paths with hard constraints
US20060015830A1 (en) * 2004-07-14 2006-01-19 Winfried Schwarzmann Systems, methods, and articles of manufacture for flexible path optimization
CN1741052A (zh) * 2005-09-22 2006-03-01 上海交通大学 车辆配载及路径优化系统
CN101055629A (zh) * 2006-04-11 2007-10-17 上海宝钢冶金建设公司 适于满载配送的车辆调度方法
CN101673382A (zh) * 2009-10-21 2010-03-17 北京交通大学 一种农资连锁经营物流配送配载组合优化方法
CN202383745U (zh) * 2011-07-13 2012-08-15 惠州天缘电子有限公司 物流信息服务系统

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
安延: "农林产品销售物流网络的构建及任务指派模型研究", 《中南林业科技大学学报》 *

Cited By (18)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105589872A (zh) * 2014-10-23 2016-05-18 阿里巴巴集团控股有限公司 图搜索结果获取方法及装置
CN105589872B (zh) * 2014-10-23 2019-01-22 阿里巴巴集团控股有限公司 图搜索结果获取方法及装置
CN106339831B (zh) * 2015-07-09 2022-04-05 菜鸟智能物流控股有限公司 用于为业务获取有效路径的方法及装置
CN106339831A (zh) * 2015-07-09 2017-01-18 阿里巴巴集团控股有限公司 用于为业务获取有效路径的方法及装置
CN105205637A (zh) * 2015-09-24 2015-12-30 天津维金斯环保科技有限公司 一种物流管理系统及管理方法
CN106919998A (zh) * 2015-12-28 2017-07-04 航天信息股份有限公司 一种基于aoe网络的物流模型及方法
CN108022060B (zh) * 2016-10-31 2022-08-16 菜鸟智能物流控股有限公司 物流网络的负载参数优化方法及装置
CN108022060A (zh) * 2016-10-31 2018-05-11 菜鸟智能物流控股有限公司 物流网络的负载参数优化方法及装置
CN109408921B (zh) * 2018-10-12 2019-09-17 西南交通大学 基于物流路径网络的复杂离散制造系统物流建模仿真方法
CN109408921A (zh) * 2018-10-12 2019-03-01 西南交通大学 基于物流路径网络的复杂离散制造系统物流建模仿真方法
CN111445093A (zh) * 2020-04-23 2020-07-24 长春工程学院 一种输电线路应急抢修路径优化系统及方法
CN111445093B (zh) * 2020-04-23 2023-04-07 长春工程学院 一种输电线路应急抢修路径优化系统及方法
WO2022018552A1 (en) * 2020-07-24 2022-01-27 International Business Machines Corporation Automated generation of optimization model for system-wide plant optimization
CN112133443A (zh) * 2020-08-25 2020-12-25 上海大学 一种基于随机游走的时序网络免疫方法
CN112133443B (zh) * 2020-08-25 2024-03-22 上海大学 一种基于随机游走的时序网络免疫方法
CN115660244A (zh) * 2022-12-27 2023-01-31 北京京东振世信息技术有限公司 路线信息生成方法、装置、设备和介质
CN116109023A (zh) * 2023-02-07 2023-05-12 广东工业大学 一种针对邮差派件的交通路径优化方法、装置及存储介质
CN116109023B (zh) * 2023-02-07 2023-09-19 广东工业大学 一种针对邮差派件的交通路径优化方法、装置及存储介质

Also Published As

Publication number Publication date
CN103399868B (zh) 2016-08-10

Similar Documents

Publication Publication Date Title
CN103399868A (zh) 一种外贸物流路径的优化方法
Saputra et al. Energy demand prediction with federated learning for electric vehicle networks
Eiselt et al. Integer programming and network models
Goh et al. Wind energy assessment considering wind speed correlation in Malaysia
CN103927643A (zh) 一种大规模订单处理与配送路径优化的方法
Deng et al. Solving a Closed‐Loop Location‐Inventory‐Routing Problem with Mixed Quality Defects Returns in E‐Commerce by Hybrid Ant Colony Optimization Algorithm
CN104050242A (zh) 基于最大信息系数的特征选择、分类方法及其装置
CN110942271B (zh) 一种基于神经网络遗传算法的智能运输调度管理方法
CN110689165A (zh) 一种基于路网可达性的车辆路径优化方法
Stopka et al. Using the operations research methods to address distribution tasks at a city logistics scale
CN111626311B (zh) 一种异构图数据处理方法和装置
CN104751250A (zh) 一种基于蚁群方法的最优路径寻找方法
CN102298737A (zh) 面向客户承诺的订单管理系统和订单管理方法
CN114111830A (zh) 一种基于ai模型的路径规划方法及装置
CN107577786A (zh) 一种基于联合聚类的矩阵分解推荐方法
CN103927628A (zh) 面向客户承诺的订单管理系统和订单管理方法
Tajik et al. A hybrid fuzzy MCDM approach for sustainable third-party reverse logistics provider selection
Alijla et al. An ensemble of intelligent water drop algorithms and its application to optimization problems
CN102799674A (zh) 面向大规模不确定物流网络的需求概率查询方法
Bougdira et al. Towards an intelligent traceability system
Mou et al. A novel ant colony system for solving the one-commodity traveling salesman problem with selective pickup and delivery
Wang et al. Using incremental clustering technique in collaborative filtering data update
Guo et al. [Retracted] Transportation Path Optimization of Modern Logistics Distribution considering Hybrid Tabu Search Algorithm
CN107169602A (zh) Agv系统设备协同优化方法
CN105550161A (zh) 一种异构系统的并行逻辑回归方法与系统

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant