基于个性化推荐的SDN企业合作伙伴选择研究

VIP免费

3.0 牛悦 2024-11-07 4 4 1.96MB 66 页 15积分

侵权投诉

摘要

随着经济全球化的迅速发展，生产要素等经济技术资源在全球范围内获得了

自由流动和优化配置，在各国分享着资源最佳配置所带来的收益的同时，也使得

各国的市场受到了不同程度的冲击。在这样的环境下，“将竞争战略向合作战略转

变”和尽量“合作共赢”将被越来越多的人们接受，多功能开放型供需网（Supply

and Demand Network with Multi-functional and Opening Characteristics for Enterprise,

SDN）的概念由此而生。

多功能开放型供需网是指以全球资源获取、全球制造、全球销售为目标，相

关企业之间由于“供需流”的交互作用而形成的一种多功能的、开放式的、供需

动态网络结构。在传统的供应链模式下，涌现出了很多合作伙伴选择和评价的方

法，对供应链企业的合作伙伴选择提供了很好的借鉴意义。SDN 作为一个充分开

放的复杂巨系统，在其供需活动的进行中，会导致海量数据的产生，这些数据不

仅在数量上表现出呈指数级增长，并且数据的更新和变换速度很快，这些新特征

决定了 SDN 合作伙伴的选择将与传统供应链企业有所不同，需要寻找新的适合供

需网特征的方法来进行合作伙伴的选择，从而提高 SDN 企业合作伙伴选择的速度

和效率。

本文首先详细分析了 SDN 企业与传统企业合作模式的不同，得出 SDN 企业

合作伙伴选择与传统企业合作伙伴选择的不同和新的特点；然后把个性化推荐的

思想引入供需网合作伙伴选择中，提出了“SDN 节点企业—待选合作伙伴企业—

合作伙伴的评价指标”的三元关系，构建了基于标签的 SDN 合作伙伴个性化推荐

系统模型，并探讨了个性化推荐的思想在 SDN 合作伙伴选择中的商业价值。其次，

提出了基于标签的 SDN 企业合作伙伴选择模型，并在个性化推荐技术研究的基础

上，分析了标签系统的思想对 SDN 企业合作伙伴选择的作用，提出了用支持向量

机算法对合作伙伴进行分类的方法，并验证了该方法的合理性。最后，以某半导

体制造企业为背景，利用支持向量机算法对合作伙伴做出分类，在分类的基础上

利用交叉验证法、遗传算法和粒子群算法对 SVM 算法中的参数进行优化，并对这

三种方法优化的结果进行比较，找出优化效果最好的参数来构建模型。

本人在对 SDN 企业和传统企业合作伙伴选择方法进行分析的基础上，采用个

性化推荐的思想推进 SDN 企业合作伙伴的选择，使其能够提高合作伙伴选择的准

确度和效率；结合企业的实际背景，采用支持向量机分类算法来建立合作伙伴选

择的模型，并利用交叉验证法、遗传算法和粒子群算法对模型参数进行优化，得

到更优的参数来构建该分类模型。

关键字：SDN 个性化推荐支持向量机遗传算法粒子群算法

ABSTRACT

With the rapid development of economic globalization, the economic and technical

resources such as the production elements flow free and allocate optimizedly in the

global scope, countries from the world share benefits brought by allocating the

resources best, in the meantime, the markets of these countries have shocked in different

degrees. In such environment, “shift the competition strategy to cooperation strategy”

and try to “win-win cooperation” will be accepted by more and more people. The

concept of Supply and Demand Network with Multi-functional and Opening

Characteristics for Enterprise has been created.

It is supply and demand network with multi-functional and opening characteristics

for enterprise that meet different supply and demand requirements through the

integration of related businesses "core competencies", under the interaction of supply

and demand flows, targeted by global resources, global manufacturing and global sales.

It emerged a lot of partner selection and evaluation methods in the traditional supply

chain model. It will produce massive data in the supply and demand activities of SDN

enterprises. These data not only grows in a series level, but also update and change

quickly, which determines the SDN partner selection will be different from the supply

chain enterprise partner selection, need to find new methods for partner selection which

is suitable for the characteristics of supply and demand network, so as to improve the

speed and efficiency of SDN partner selection.

Firstly, differences of SDN and traditional enterprises cooperation patterns are

analyzed, the differences and new features of SDN enterprises partner selection is

proposed. The technology of personalized recommendation is mixed into the SDN

enterprises partner selection, the three ternary relationships of “SDN node enterprise –

partners’ enterprise to choose – partner’s evaluation index” is set up, label-based

partners personalized recommendation system model for SDN is established, and the

commercial value of the personalized recommendation thoughts for SDN partner

selection is discussed. Secondly, based on the study of personalized recommendation

technology, this paper analyzes the label system’s effect for the SDN enterprise partner

selection, the method of using support vector machine to select the SDN enterprise

partner is put forward, and the rationality of this method is verified. Finally, as a

semiconductor manufacturing enterprise as example, the relationship about strong

correlation or weak on of SDN between supply and demand partner will be forecast

using support vector machine algotithm, then cross validation method, genetic

algorithm and particle swarm algorithm are used to optimize the parameters of SVM

algorithm, and the results of these three methods are compared to find out the

parameters which have the best effect, uses these parameters to build the SVM model.

Based on the analysis of selection methods for SDN enterprises and traditional

enterprises, the thought of personalized recommendation is using to propel SDN

enterprises partners selection, which will improve accuracy and efficiency of partner

selection. Combined with enterprise’s actual background, the relationship between

strong-correlated and weak-correlated partners of SDN is predicted using SVM

classification algorithm, and cross validation method, genetic algorithm and particle

swarm algorithm are used to optimize the model parameters, and get better parameters

to build the classification model.

Key Words: SDN, personalized recommendation, Support Vector

Machines (SVM), Genetic Algorithm, Particle Swarm Algorithm

中文摘要

ABSTRACT

第一章绪论 ···················································································································· 1

1.1 论文的研究背景和意义 ························································································ 1

1.2 国内外研究现状 ··································································································· 4

1.2.1 SDN 的研究现状 ··························································································· 4

1.2.2 个性化推荐的研究现状 ·············································································· 7

1.3 论文的研究内容 ··································································································· 9

1.4 论文的研究思路及创新点 ················································································· 10

第二章支持向量机和智能算法相关理论 ·································································· 12

2.1 支持向量机算法的基本理论 ············································································· 12

2.1.1 支持向量机的概念 ····················································································· 12

2.1.3 支持向量机分类 ························································································ 14

2.1.4 支持向量模型的特点 ················································································ 17

2.2 遗传算法和粒子群算法基本理论 ····································································· 17

2.2.1 遗传算法基本理论 ···················································································· 17

2.2.2 粒子群算法基本理论 ················································································ 18

2.2.3 遗传算法与粒子群算法的比较 ································································ 19

第三章 SDN 节点企业合作伙伴选择研究 ·································································· 21

3.1 SDN 节点企业合作伙伴选择方法 ······································································ 21

3.1.1 传统供应链企业合作伙伴的选择方法 ····················································· 21

3.1.2 供需网企业与传统企业合作伙伴选择比较 ·············································· 23

3.1.3 SDN 企业合作伙伴选择的独特性 ······························································ 25

3.1.4 SDN 企业合作伙伴选择分析 ···································································· 28

3.2 基于支持向量机的 SDN 企业合作伙伴选择 ··················································· 31

3.2.1 支持向量机算法用于 SDN 企业合作伙伴选择 ······································· 31

3.2.2 交叉验证法对合作伙伴选择的优化 ·························································· 32

3.2.3 遗传算法和粒子群算法对合作伙伴选择的优化 ······································ 33

第四章基于个性化推荐的 SDN 合作伙伴选择模型 ················································ 36

4.1 用个性化推荐技术推进 SDN 合作伙伴决策 ··················································· 36

4.1.1 个性化推荐技术对数据的智能化处理 ····················································· 36

4.1.2 个性化推荐支持 SDN 合作伙伴选择 ························································· 37

4.1.3 标签系统的思想在 SDN 节点企业合作伙伴选择中的应用研究 ··········· 39

4.1.4 个性化推荐为 SDN 企业带来的价值 ························································· 41

4.2 基于标签系统的 SDN 企业决策模型 ································································· 41

4.2.1 SDN 企业三维结构模型的构建 ·································································· 41

4.2.2 基于标签的 SDN 节点企业合作伙伴选择系统框架 ································· 42

4.2.3 基于标签的 SDN 节点间的协同决策 ························································· 45

第五章基于 SVM 的SDN 合作伙伴选择的实验分析 ············································· 47

5.1 某半导体制造企业的背景资料 ······································································· 47

5.2 实验分析与结果 ································································································· 47

5.2.1 数据的收集与预处理 ················································································· 47

5.2.2 支持向量机算法的实现与分析 ································································· 48

5.2.3 交叉验证法优化分类器参数 ···································································· 49

5.2.4 遗传算法和粒子群算法优化分类器参数 ················································ 51

5.2.5 优化结果的比较和分析 ············································································ 53

第六章总结与展望 ······································································································ 54

6.1 全文工作总结 ···································································································· 54

6.2 进一步研究展望 ································································································ 54

参考文献 ························································································································ 56

在读期间公开发表的论文和承担科研项目及取得成果 ············································ 59

致谢 ································································································································ 60

附录 ································································································································ 61

第一章绪论

1.1 论文的研究背景和意义

随着经济全球化的迅速发展，生产经营活动用到的各种资源在全球范围中得

到了广泛的流通和分配，在各国分享着资源最佳配置所带来的收益的同时，也使

得各国的市场受到了不同程度的冲击。很多不确定的因素涌现出来，市场环境的

瞬息万变、现代科技日新月异的发展、客户需求的个性化和多样化、以及客户对

产品各方面的更高要求，这些都使企业面临着前所未有的挑战。然而企业传统的

管理理念和模式已无法适应快速变化的市场环境，全球资源的自由流动和优化配

置迫使企业进行管理理念方面的转变，供应链内的完全竞争和战略联盟、企业集

群、虚拟企业等模式的内部合作、外部竞争的观念都将被淘汰，企业高层需要对

公司战略进行调整，与人分享和合作的理念将更多的被人们所接受。

在该形势下，徐福缘教授经过深入的研究给出了多功能开放型企业供需网的

概念。多功能开放型企业供需网（Supply and Demand Network with Multi-functional

and Opening Characteristics for Enterprise, SDN）是指以全球资源获取、全球制造、

全球销售为目标，企业之间根据自身的需求关系而形成了一种多功能的、开放式

的、供需动态网络结构[1]。SDN 是适应全球经济一体化时代的一种基于充分合作

与共赢的新的管理理念和模式。

SDN 作为对传统供应链管理模式的一种改进，克服了传统供应链的链状结构、

功能单一和随时可能发生断链的不足，为全球企业的充分合作和共赢创造了有利

的条件，促使资源的全球流通，用开阔的视野，从全球范围寻找更优质更廉价的

原材料，达到物尽其用、人尽其才，生产出可以在全球范围销售的产品和服务，

从而促使全球进出口贸易的发展，以低耗材、低浪费、低污染实现全球经济和谐

稳定的可持续发展。

供需网具有如此多的优势，将促使公司和个人努力推进供需网的形成，以改

善传统供应链所存在的问题。供需网的形成需要一个过程，在 SDN 的IDEF 模型

中，SDN 的构建包括以下五个方面：市场机遇识别，动态合作子网目标的确定，

伙伴核心能力识别、伙伴的选择和优化，组织与运行模式的选择，运行与反馈[1]。

作为供需网构建的一个过程，合作伙伴的选择与优化是 SDN 企业普遍重视的一个

阶段，它决定着 SDN 的构建是否成功。

传统供应链企业在选择合作伙伴时用到了很多种方法，这些方法经过了很多学

者的研究和改进，对供需网企业合作伙伴的选择有很好的指导和借鉴意义。然而

SDN 作为对传统供应链的一种改进形式，在合作伙伴的选择上必将呈现出新的特

点，伴随着信息技术的发展以及供需网开放性的特点，SDN 企业在其正常的工作

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

15 积分 4人已下载

立即下载 VIP免费下载

摘要：

摘要随着经济全球化的迅速发展，生产要素等经济技术资源在全球范围内获得了自由流动和优化配置，在各国分享着资源最佳配置所带来的收益的同时，也使得各国的市场受到了不同程度的冲击。在这样的环境下，“将竞争战略向合作战略转变”和尽量“合作共赢”将被越来越多的人们接受，多功能开放型供需网（SupplyandDemandNetworkwithMulti-functionalandOpeningCharacteristicsforEnterprise,SDN）的概念由此而生。多功能开放型供需网是指以全球资源获取、全球制造、全球销售为目标，相关企业之间由于“供需流”的交互作用而形成的一种多功能的、开放式的、供需...

展开>> 收起<<

基于个性化推荐的SDN企业合作伙伴选择研究.pdf

共66页,预览7页

还剩页未读，继续阅读

基于个性化推荐的SDN企业合作伙伴选择研究

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

推荐作者

热门标签

举报选择: