【作者】 钟瑞琼；

【导师】 罗伟其；

【作者基本信息】 暨南大学 ， 管理科学与工程 信息管理与信息系统， 2021， 博士

【摘要】 我国产学研合作创新网络存在科技成果转化率、投入产出比远低于美国等发达国家等问题,资源配置不合理成为阻碍我国产学研合作向纵深发展的关键。现有研究忽视了合作创新网络的动态性、非线性特征,并且无法对人员和资金进行精准预测与控制。针对现有研究的不足,论文通过建立系统动力学模型研究合作创新网络的动态运行机制;基于反馈控制理论,构建产学研合作创新网络前馈-反馈跟踪控制模型实现精准预测与控制,为优化创新资源配置提供新的理论模型和方法支撑。研究结果表明:第一,创新投入、网络中心性、网络联系密度和网络开放性均对创新产出有显著的正向作用,知识转移具有正向调节作用;第二,创新投入对创新产出的间接影响均为显著的正向作用,并通过网络结构这一中介机制产生正向影响,其中网络开放性、网络联系密度变量在对专利数量影响效应中起到了中介作用;第三,所提出的前馈-反馈控制模型和小样本预测方法能实现精准预测与动态控制,并且PDM＿BOA-SVR＿GS-MLP(粒子动态多阶段扰动＿蝴蝶优化-支持向量回归＿网格搜索-多层感知机)模型为最优控制模型。本文的理论和实证研究成果,有助于完善产学研合作创新网络体系理论,在一定程度上丰富和拓展了反馈控制领域的研究。更多还原

【Abstract】 The transformation rate of scientific and technological achievements and the input-output ratio of my country’s industry-university-research cooperation innovation network are far lower than those of developed countries such as the United States.The unreasonable allocation of resources has become the key to hindering the in-depth development of my country’s industryuniversity-research cooperation.Existing research ignores the dynamic and nonlinear characteristics of cooperative innovation networks,and cannot accurately predict and control personnel and funds.In view of the shortcomings of existing research,the paper studies the dynamic operation mechanism of the cooperative innovation network by establishing a system dynamics model;based on the feedback control theory,a feedforward-feedback tracking control model of the industry-university-research cooperative innovation network is constructed to achieve accurate prediction and control.It provides new theoretical models and method support for optimizing innovation resource allocation.The research results mainly include three aspects.Firstly,innovation input,network centrality,network connection density and network openness all have significant positive effects on innovation output,and knowledge transfer has a positive moderating effect.Secondly,the indirect impact of innovation input on innovation output is a significant positive effect,and the effect is achieved through the intermediary mechanism of network structure,in which the variables of network openness and network connection density play an intermediary role in the impact on the number of patents.Thirdly,the proposed feedforward-feedback control model and small sample prediction method can achieve accurate prediction and dynamic control,and PDM＿BOA-SVR＿GS-MLP(Particle Dynamic Multi-Stage Perturbation＿Butterfly Optimization-Support Vector Regression＿Grid Search-Multiple layer perceptron)model is the optimal control model.The theoretical and empirical research results of this paper are helpful to improve the theory of the industry-university-research cooperation innovation network system,and to a certain extent enrich and expand the research in the field of feedback control.更多还原