Research On Robustness Optimization Of Supply Chain Network Based On Multi-agent

With the rapid development of economic globalization,the complexity of supply chain is gradually increasing,and the chain structure has evolved into a network structure,forming a supply chain network.At the same time,all kinds of emergencies occur frequently,resulting in the enterprises in the supply chain network cannot supply normally,thus affecting the normal supply-demand relationship between upstream or downstream businesses.And the enterprises cannot operate normally due to lack of materials,resulting in the proliferation of failure in the whole supply chain network,resulting in the cascading failure of the whole supply chain network,Finally,the productiveness of the network is seriously reduced,even paralyzed,resulting in huge economic losses.Therefore,the research on how to enhance the resistance of supply chain network to interruption and improve the robustness of network is of great value and significance.Based on the complex network theory,this paper takes the empirical supply chain network as the object,and uses the multi-agent modeling technology to explore the robustness optimization method of supply chain network from network structure and cascading failure.The main research work is as follows:(1)Structural robustness optimization of supply chain network based on improved degree preserving cross rewiring.Aiming at the problem of insufficient antiinterference ability of empirical supply chain network structure,starting from the network topology,in order to improve the robustness of network structure,optimize the topology of supply chain network.Firstly,based on the complex network theory,according to the empirical data,the empirical supply chain network is constructed,and its robustness is analyzed.Considering the internal attributes of enterprises in the realworld supply chain network,the concept of node fitness is introduced,and a structural robustness optimization method which based on fitness preserving cross rewiring is put forward.Finally,the multi-agent simulation platform is built based on the multi-agent simulation tool repast,and the corresponding simulation experiments are designed to compare and explore the robustness optimization effect of the improved rewiring method and the traditional rewiring method,thus verifying the effectiveness of the method.(2)Research on node importance identification and protection method considering cascading failure.Aiming at the problem that the interruption of key enterprises is more likely to cause cascading failure and significantly reduce the network connectivity of supply chain,the node importance evaluation research before interruption is carried out.Through the formal description and modeling of cascading failure of supply chain network,a node importance evaluation method based on network load entropy is proposed to identify and maintain important enterprises in supply chain network,so as to improve the robustness of supply chain network.Finally,according to the multiagent simulation platform,in order to verify the validity of our proposed evaluation method,the comparative experiments are designed.And the corresponding key node protection methods are proposed.(3)Research on supply chain network recovery method considering cascading failure.In view of the problem that it is difficult to prevent the supply chain network from being affected by emergencies and even lead to large-scale network paralysis only through prior prevention,research on the recovery method of supply chain network after interruption is carried out to enhance the robustness of the network against cascading failure by improving the recovery ability of the network.A target recovery method based on node importance is proposed,a cascade failure model with recovery method is established,and the node overload failure probability is introduced into the model.Finally,according to the multi-agent simulation platform,the model is used to compare and explore the recovery effect of the proposed target recovery methods and the optimization effect of different recovery methods on network robustness.