【作者】 李星；

【导师】 范如国；

【作者基本信息】 武汉大学 ， 管理科学与工程， 2011， 博士

【摘要】 在绿色约束和低碳经济环境下,传统集群的劣势和非可持续性发展问题日益显现出来,传统集群如果不能按照绿色约束和低碳经济的要求尽快实现创新,将会很快被淘汰。而且随着中国逐步迈向中等收入社会,国内市场的扩张成为未来经济增长的主要动力,需要更多的本土企业通过创新,创造出服务于中国市场、适合中国国情的新产品、新服务以及新的经营组织方式。而在复杂的国内外市场竞争面前,企业的个体性创新有时并不能应对这些挑战,此时需要集群企业与集群网络内的其它行为主体、以及与网络外部的各行为主体之间进行多方位、多层次的合作,以达到资源共享和相互学习,从而拓展集群创新的空间,获得提高集群企业创新能力而不具备的互补性资源,因而集群企业之间的创新演变成了一种网络化的创新。20世纪80年代以来,创新过程中企业之间以及个人之间的联系所形成的网络受到人们的高度重视,经济学、社会学以及产业组织学等学科从不同角度对创新网络进行了研究。随着对企业集群研究地不断深入,理论界的研究视角开始从集聚的规模效应转移到企业集群内部创新活动。然而,现有文献对集群创新网络中合作问题的研究,并没有从真正意义上来考虑网络的结构特性对合作过程中一系列行为产生的影响,社会网络分析的工具有很多,包括用图论方法分析网络的拓扑关系,到建立严密的数量模型,引入很多数学工具,并结合了仿真试验,它已经从一种具体的研究方法拓展为一种理论框架。因此,本文从社会网络的角度出发,结合演化博弈理论、概率论、系统科学以及Opinion动力学的相关理论对集群创新网络多主体间的合作机理来进行研究,研究内容主要包括以下几个方面：(1)对企业集群创新网络内多行为主体及其链接模式(包含核心网络、支撑网络、外部网络)进行了分析,并探讨了核心网络层主体间、核心网络层主体与支撑网络层主体之间、以及核心网络层与外部网络层的互动机制,并在此基础上,从三个方面论述了集群创新网络中多主体间合作的动力因素。另外,由于集群这种俱乐部产品的一个重要特性就是集群创新网络内的企业数目增加到一定程度时,会产生网络的拥挤效应,而且随着集群创新网络外部新企业的进入,集群公共产品的边际收益也会呈现一种递减趋势,因而对集群创新网络最优边界的确定问题也进行了研究。(2)在阐述企业集群创新网络结构特点的基础上,分析了创新网络内各行为主体的决策行为模式、群体(或社团结构)对创新网络内行为主体的影响函数,提出了在群体(或社团结构)影响下,创新网络内主体行为调整的结构模型,并引入了正面权重影响指数和负面权重影响指数来衡量群体(或社团结构)对创新网络内行为主体的影响程度,在此基础了,利用Opinion动力学对创新网络中主体决策行为相互影响下的合作涌现机制进行了分析,包括模型的构建与仿真结果的分析。(3)建立了一个企业集群创新网络内多企业动态合作创新微分博弈模型,在给定两个假设前提下,结合实际案例从主体理性的角度比较分析了在合作创新与自主创新情形下集群创新网络内企业所采取的最优策略及其对应的最优价值函数：一是企业的投资策略对企业技术水平影响；二是技术水平的变化对企业价值函数的影响；三是企业投资所获得的价值对企业未来投资策略的影响；研究表明参与合作创新的企业利用集群创新网络内的资源云,通过较少资本的投入,就可以带来企业技术水平的改进,而企业技术水平的改善可以更快地为企业创造更多的价值,并且这种价值的增长对企业未来的投资策略产生了积极的影响,从而可以有效地防止集群的衰退。同时,本文选取54个国际级高新产业技术开发区作为实证样对合作创新能力与合作创新绩效进行了分析。(4)利用概率论中的状态转移方程对集群创新网络多主体合作信任机制的动态性进行了分析,并利用仿真手段分析了企业集群创新网络中企业间的信任关系的动态演变特性,得出了与集群发展实际相吻合的四个重要结论。同时,利用仿真手段分析了集群创新网络的结构特性对网络内企业间信任关系状态的影响：一是网络中企业的中心度(即企业的影响力)；二是网络中企业间的路径长度(即地理位置的接近性)；三是网络中企业间的强弱连接；四是网络中企业的结构自治度；也得出了与实际相吻合的四个重要结论,并通过一个实际案例对集群创新网络多主体合作信任机制的演化模型进行了验证。更多还原

【Abstract】 Under the economical environment of green New Deal and low-carbon, the problem of traditional disadvantages of the cluster and non-sustainable development is becoming increasingly apparent, the traditional cluster will soon be eliminated if it won’t innovate in accordance as soon as possiblewith the acquirement of green New Deal and low-carbon economy. As China entering the middle-income community, the expansion of the domestic market becomes the main driver of future economic growth, and it needs more and more local enterprises to create new product, new services and new business organization to service the market in China and suitable for China’s condition through innovation. However, in face of the complex domestic and international market competition, individual enterprises’innovation sometimes can not meet these challenges, so the enterprise in the industrial cluster now needs the various multi-faceted, multi-level cooperation with the actors in the industrial cluster network and other actors outside the industrial cluster network to share resource and learn from each other, so it can expand the space of industrial cluster innovation and get the complementary resource to improve the enterprises’ innovation ability in industrial cluster. Therefore, the innovation among enterprises in industrial cluster has evolved to the innovation in way of network.Since 80 years of the 20th century, people has attached great importance in the network formed by links among enterprises and individuals in the process of innovation, economics, sociology and other disciplines of Industrial Organization studied the innovation networks from a different perspective. Theorists’research perspective shifted to the innovation activities within the SME cluster from the scale effect of gathering with their depth research on the SME cluster. However, the existing literatures’reaserach on cooperation in cluster innovation network didn’t consider the impact of the structure features of the network on a series of acts in the process of cooperation, there are many social network analysis tools, including the use of graph theory to analyze the network topology, establishing strict models, the introduction of many mathematical tools, and combined with a simulation test, so it has expanded as a theoretical framework from a specific research method.Therefore, Multi-agents’ cooperation mechanism in cluster innovation network was studied in this article from the perspective of social networks, combining with the theory of evolutionary game theory, probability theory, system science and Opinion Dynamics, and the research content is as follows:(1) Multi-agents and their link model in cluster innovation network (including the core network, supporting network, external network) were analyzed and discussed the interaction mechanism among agents in the core network layer, between the agents in the core network layer and the agents in the supporting network layer, as well as between the core network layer and the external network layer were also discussed in this paper, and then we explained the cooperation dynamic factors among Multi-agents in the cluster innovation network from three aspects. In addition, an important feature of the cluster as club product is that it will have a network crowding effect when the number of enterprises in the cluster innovation network extended to certain degree, and the marginal benefit of cluster as public goods will show a decreasing trend as the entry of new firms outside the network, therefore, the determination of optimal boundary of the cluster innovation network were also studied in this paper.(2) On the basis of describing the structural features of enterprise cluster innovation network, we analysed behaviour decision-making model of agents in the innovation network, the impact function of the group(or community structure) on the agents in the innovation network, presented the structure model of adjusting agents’ behaviour in the innovation network under the effect of the groups(or community structure), introduced the positive weight index and the negative weight index to measure the influence degree of group(orcommunity structure) on the agents in the innovation network, and then we use Opinion dynamics of theoretical physics to analyse the cooperation emerging mechanism under the mutual influence of agents’ decision-making behaviour, including constructing the model and analysing the simulation results.(3) The differential game model of Multi-enterprises dynamic cooperation innovation in the cluster innovation network was established in this paper, given the two assumptions, we compared and analysed the optimal strategy adopted by enterprises in the cluster innovation network and the corresponding optimal value function in case of cooperation innovation and independent innovation from the perspective of agents’ reason combining the actual case:Firstly, the influence of enterprises’ investment strategy on the enterprises’ technology level; Secondly, the impact of change of enterprises’ technology level on the enterprises’ value function; Thirdly, the impact of enterprises’ investment value obtained on the company’s future investment strategy; research shows that companies participating in cooperation innovation can improve the technology level with less capital investment through using the resources cloud within the cluster innovation network, and the improvement of technology can create more value to the company, and this growth of the value had a positive impact on company’s future investment strategy, so it can prevent the decline of cluster effectively. At the same time, cooperation innovation performance is also analysed from the perspective of empirical in this paper.(4) the dynamic feature of trust mechanism of Multi-agent cooperation in the cluster innovation network was analysed by using state transition equation in probability theory, and the dynamic evolution characteristics of trust relationship among enterprises in the cluster innovation network was also analysed by using simulation tools, and we obtained four important conclusions which is consistent with the development of actual cluster. At the same time, we analysed the impact of the structural characteristics of the cluster innovation network on enterprises’ trust relationship state in the cluster innovation network through simulation:Firstly, the enterprises’ centrality in the innovation network(the enterprises’ impact); Secondly, the path length among enterprises in the innovation network (the geographical proximity); Thirdly, the connection strength among enterprises in the innovation network; Fourthly, the self-government degree of enterprises’ structure; and we also obtained four important conclusions which is consistent with the actual case, and we validated evolutionary model trust mechanism of Multi-agent cooperation in the cluster innovation network through an actual case.更多还原