Abstract:

High-end equipment is characterized by high technological complexity and significant reliance on critical external components， making enterprises in this sector particularly vulnerable to multi-source risks such as technological blockades， trade sanctions， and geopolitical conflicts in unstable and uncertain environments. These risks often overlap， propagate， and amplify， leading to cascading disruptions， operational instability， and substantial losses. Constructing multi-source risk transmission networks and identifying critical nodes and paths within these networks is essential for effectively preventing and mitigating risks in high-end equipment manufacturing.In this study， a systematic methodology is developed to construct multi-source risk transmission networks and identify critical nodes and paths. Potential risk factors are identified using phrase extraction techniques combined with the BERTopic model. Semantic association rules between risk factors are defined， and their associations are determined using a sliding window similarity method， which facilitated the construction of a multi-source risk factor association network. The intrinsic risk strength of each risk factor， as well as the transmission strength between factors， is quantified to transform the association network into a multi-source risk transmission network. Weighted closeness centrality is employed to measure the transmission strength of risk factors in the network， and critical nodes are identified by integrating their intrinsic risk strength with transmission potential. Finally， the Choquet fuzzy integral is applied to non-linearly aggregate multi-source risks， and an ant colony algorithm is exploited to determine critical transmission paths within the network.Risk texts are collected through a mixed-method approach： online texts are crawled from prospectuses and annual reports of listed high-end equipment manufacturers， while offline texts are gathered through on-site interviews with managers and employees at a CRRC subsidiary to supplement operational risk-related narratives.Key findings include　1） Identification of 91 risk factors and their associations， forming a multi-source risk factor association network； 2） Derivation of 28 two-source concurrent risk scenarios from eight primary risk sources， leading to the construction of 28 two-source risk transmission networks； 3） Frequent identification of critical risk factors such as insufficient production capacity， rising production costs， and declining market demand in two-source risk transmission networks， alongside other critical factors like damaged production equipment， reduced industrial investment， and reliance on imported raw materials； 4） Observation that two-source risks tend to converge and amplify at critical factors such as insufficient production capacity， rising production costs， and declining market demand； 5） The most detrimental critical paths emerged from the concurrent occurrence of public health emergencies and macroeconomic downturns， causing peak-level risks and severe operational damage to enterprises.

Key words: high-end equipment manufacturing, multi-source risk transmission networks, critical nodes, critical paths