Short-time Passenger Flow Prediction Model based on Combined Model for Large Events in and Out of Rail Stations
- Resource Type
- Conference
- Authors
- Shi, Pingcui; Hu, Hua
- Source
- 2022 4th International Conference on Artificial Intelligence and Advanced Manufacturing (AIAM) AIAM Artificial Intelligence and Advanced Manufacturing (AIAM), 2022 4th International Conference on. :208-212 Oct, 2022
- Subject
- Computing and Processing
Rails
Thermal factors
Organizations
Predictive models
Market research
Data models
Kalman filters
rail stations
short-time passenger flow prediction for large events
Unscented Kalman filter
wo Layer Long short-term memory
attention mechanisms
- Language
Addressing the difficulties of acyclicity, strong abruptness and little historical passenger flow data in predicting short-time passenger flow in and out of rail stations during large events, an UKF-AMLSTM-based short-time passenger flow prediction model is established. Considering large event factors and other external factors, unique thermal coding is performed and then used as input to the Two Layer Long Short Term Memory Network (LSTM) along with the historical passenger flow data after flat Unscented Kalman filter (UKF) slip processing, with Attention Mechanism (AM) added to identify key input steps. The validity of the model is verified by taking Shanghai Metro Xujing East Station as an example. The research results can provide a basis for the development and optimization of passenger organization schemes in rail stations during large events.