The Industrial Complex Networks' Modeling, Simulation And Analysis

The industry complex network in the paper include the industry competition complex net-work and the emerging collaboration network based on Peer Production. In fact, enterprisesexist in large networks formed by many competitors. Industry competition complex networkmodel and the analytical method complement the classical Organization Analysis theory fromnetwork perspective among the enterprises, which develops the industrial organization theoryand method. However,most industry complex network analysis focuses on competition net-work, the paper will both study the industry competition complex network and Peer Productionnetwork.In addition, the research on industry complex network mostly focus on single industryand lack general analytical framework and methods to all industries. First of all, through theanalysis of many specific real competition network and Peer Production network, the paperconstruct a universal framework of industry competition complex network for"static topologycharacteristics—subgraph characteristics—generating mechanism—dynamic characteristics".Second, we make deep study on the static characteristics of the network topology, mainlyincluding the application of maximum likelihood method to estimate the degree distribution ofindustry complex networks, investigating the static structure especially the subgraph charac-teristic of industry complex network by the analysis of k-clique subgraph, motif, hierarchicalstructure, community structure, fractal and multifractal property. We propose the motif dis-assortative model, the imitated Yule process model and the bipartite fitness model to analyzethe generating mechanism of complex network. Again, we make research on the dynamicproperty of the industry competition network and Peer Production collaboration network. Weconstruct the competition spread model over the industry competition network and find thatThe enterprise location in the network topology who launch the competition have a significanteffect on the spread result. We also find that competitive effects spreading over the network haslocal characteristic. we construct and analyze the platform competition model of peer produc-tion,which reveals the evolution of peer production platform competition when the competitionintensity based on the network topology follows the power law and Gaussian distribution. Wefurther construct the information spread model based on Peer Production collaboration networkand analyze topological property of the individual who initiate the information as well as thenumber impact significantly on the spread result. Then we construct the multi-agent adaptivesystem model to study the information spread process and reveal the macro emergence of theadaptive system during different process. We also analyze the stability of Peer Production by constructing Ising model combining herding effect and study the the critical point of the sys-tem impacted by different network structures and external conditions, and finally we use MonteCarlo simulation to verify the critical analytical solution derived by mean field theory. We con-struct multi-agent self-adaptive system and analyze the stability of peer production system inshort term and long term. On the research of network dynamics, we construct the competitionspread model, Peer Production platform competition model, information spread based on PeerProduction network and multi-agent model, Ising model and multi-agent model, which are notonly"comprehensive"innovative but also more realistic, for example multi-agent model con-sider the"dead"node as a"live"agent, it bring the complex adaptive system into the analysisof industry complex network which make the industry complex network method concern theenterprises'practical autonomy. Further, During the analysis process of the static and dynamiccharacteristics of the network, we improve the corresponding algorithm. For example in theanalysis of static network topology, we improve the traditional box covering algorithm for com-plex network fractal property analysis and verify the efficiency of the the improved algorithmand in the process of dynamics analysis modeling, we improve PSO from the perspective ofstatic network topology and dynamic network topology, propose PSO based on SFL and DSF-PSO two improved models.Finally, be noted that the above framework, model and approach not only has the indus-try's complex network analysis universal significance, but also has the new insights into theindustry complex network and industry analysis. For example we adopt the maximum likeli-hood method to estimate the r_min values of the industry complex network's degree distributionwhich reveal the smallest competitors numbers of power-law distribution, and the degree fol-low the power-law distribution depict the heterogeneity within competitors of industry complexnetwork and heterogeneity within Collaborators of Peer Production network. The discoveryof subgraph, motif, community structure, fractal and multifractal property in industry complexnetwork respectively reveals the existence of a stable industrial competition and overlappingcompetition structure, with hierarchical, community characteristics, and self-similarity and soon.