MFMM: a passenger flow prediction model for rail transit systems considering multiple external environmental factors at stations

This is the authors' abstract. We don't add key points for this paper.

Accurate predictions of short-term passenger flow in subway systems can assist operators in designing train operation plans to ensure efficient operation, conserve energy, and alleviate station congestion. Many existing studies focus on model structure and neglect the impact of dynamic changes in the station's external environment on subway passenger flow. Therefore, this study proposes a subway passenger flow prediction model for a station that considers the weather conditions and geographical characteristics around the station. The model extracts dynamic weather feature perturbation factors and static geographical characteristic weighting factors for feature fusion. In addition, based on the principles of Fast Fourier Transform and MS-TimesBlock structures, the model integrates two specific modules to fuse multi-source weather data and geographical characteristic data. Finally, three representative stations are selected to test the capabilities of the proposed prediction model and the need to incorporate weather features and geographical characteristics.