System identification for designing a crowd danger controller in a metro station during large passenger flow

Abstract

High-density passenger influxes in metro hubs create significant safety risks. Traditional passenger flow control studies often adopt a macroscopic perspective, focusing on service-oriented objectives rather than mitigating microscopic crowd-related risks. This approach frequently overlooks the localized dangers at architectural bottlenecks. To address this, we present a methodology for modeling these microscopic dynamics for future control applications. First, a comprehensive microscopic simulation model is developed, operating on two levels: a tactical level for route choice and an operational level using the Social Force Model (SFM) for pedestrian movement. A practical control input, defined as a dynamic entry ticket gate service time delay, and a state-of-the-art output metric, defined as crowd danger, are formulated. The core contribution is the derivation of a linear state-space representation from this complex, non-linear simulation using system identification techniques based on generated input-output data. Validation results demonstrate that the identified linear model effectively captures the essential system dynamics and provides accurate multi-step-ahead predictions. This work validates a robust process for “distilling” complex crowd phenomena into a tractable, control-oriented model, providing a reliable foundation for the future design of model-based safety systems.