【摘要】 在制造业智能化转型中，人工智能(AI)技术的应用提高了效率但也带来了风险。现有研究缺乏对复杂的制造网络动态演化及AI双向调节作用的系统分析。本文整合复杂网络与元胞自动机方法，构建动态风险传导模型，量化AI能力对风险传播和恢复的双面效应，并通过特斯拉、西门子、富士康等案例验证策略有效性。研究发现，AI能力通过增强节点交互效率加速风险传播，同时通过智能优化提升系统恢复效率，形成“传播加速-恢复增强”的动态平衡。研究还发现，运行状态特征对风险控制的影响超过了网络结构特征，AI能力可以通过优化运行状态的稳定性来降低风险。在高AI能力的条件下，采取针对性策略的风险显著低于随机策略。研究为制造业提供了平衡AI创新与风险管控的量化模型和实践路径，建议重点提升关键节点AI韧性、实施差异化网络保护，并建立跨组织风险协同治理体系。更多还原

【Abstract】 In the intelligent transformation of the manufacturing industry, the application of Artificial Intelligence(AI)technology has improved efficiency while introducing concomitant risks.Existing research lacks a systematic analysis of the dynamic evolution of complex manufacturing networks and the bidirectional regulatory role of AI.This study integrates complex network theory with cellular automata methods to construct a dynamic risk transmission model, quantifying the dual-sided effects of AI capabilities on risk propagation and system recovery.The effectiveness of the proposed strategies is validated through case studies of leading manufacturing enterprises such as Tesla, Siemens, and Foxconn.Key findings of the research are as follows: First, AI capabilities accelerate risk propagation by enhancing the interaction efficiency of network nodes, while simultaneously improving system recovery efficiency through intelligent optimization, thereby forming a dynamic balance characterized by“propagation acceleration-recovery enhancement”.Second, the operational state characteristics of the manufacturing system exert a more significant impact on risk control than the network structure characteristics, and AI capabilities can mitigate risks by optimizing the stability of operational states.Third, under the condition of high AI capabilities, the risk level associated with targeted strategies is significantly lower than that of random strategies.This study provides the manufacturing industry with a quantitative model and practical pathway for balancing AI-driven innovation and risk management.It offers three core recommendations: prioritizing the enhancement of AI resilience in key network nodes, implementing differentiated network protection measures, and establishing a cross-organizational collaborative governance system for risk management.更多还原