突发事件下数据驱动交通网络延误最优控制方法OA

[Objective]To address the complex mechanisms of delay propagation in transportation networks and the challenges posed by significant modeling errors and high computational complexi-ty in conventional model-driven control approaches,this study proposes a data-driven delay-control framework.[Method]First,a diffusive delay-propagation model is developed,which quantifies the interactions among network nodes through state-space equations,thereby overcoming the limitations of conventional discrete-event models.Second,a data-driven control algorithm is designed to derive a closed-loop optimal control strategy by analyzing stochastic experimental data.This strategy estab-lishes a direct relationship between system inputs and outputs via pseudo-inverse operations,thereby eliminating explicit reliance on delay-propagation and control matrices,while reducing computation-al complexity.Finally,a noise-compensation mechanism is incorporated to enhance the robustness of the control strategy.[Data]Using operational data from the Belgian railway network,including train timetables,actual running times,delay records,and geospatial information,among others.[Result]The results indicate that the proposed method effectively suppresses delay propagation trig-gered by local disturbances,with delays at non-failure nodes approaching zero after control.Com-pared with conventional optimization algorithms,the proposed method exhibits lower computational complexity and more stable control performance.The total control cost increases with the control ho-rizon;however,the cost per unit time decreases,reflecting the algorithm's time-optimization capabil-ity.When the number of failed nodes exceeds a critical threshold,the network becomes uncontrolla-ble and the control strategy fails,thereby defining the effective operational boundary of the algo-rithm.[Conclusion]The proposed method is data-driven,eliminating the need for precise system modeling and reducing the impact of modeling errors.It improves applicability in practical engineer-ing scenarios and provides a theoretical foundation for real-time delay control in complex transporta-tion networks.