Abstract:Effective management of urban rail transit systems increasingly centers on short-term passenger flow prediction which is a key factor in operational scheduling decisions. The critical need to use existing infrastructure efficiently and avoid potential emergencies due to large crowd gatherings. The advancements in short-term passenger flow forecasting have become essential within the intelligent transportation system domain. Despite its importance, there is a prominent gap in research specifically addressing the prediction of various types of passenger flows in subway systems. To bridge the gap, the paper proposed an innovative integrated method combining CNN-LSTM and SVM with Wavelet Transform harnessing their collective strengths. The approach involves a multi-stage process. Initially, passenger flow data is dissected into high and low-frequency series using wavelet transform. In the prediction phase, CNN-LSTM and SVM techniques are employed to, respectively learn and forecast these high and low-frequency sequence. The final phase involves the reintegration of these various predicted sequences through wavelet transform. The integrated approach aims to provide a more accurate prediction of short-term passenger flows in urban rail transit catering to the specific demands of the complex and dynamic field. The analysis shows that the SVM based Wavelet Transform method outperforms the CNN-LSTM based Wavelet Transform method in terms of efficiency and accuracy. Both approaches achieve better results than classical prediction techniques for passenger flow prediction on evaluation metrics, i.e., MAPE (9.80% and 7.28%), VAPE (0.78% and 3.85%), RMSE (2.48% and 0.76%), MSE (14.86% and 4.65%), R 2 (1.57% and 12.57%) and accuracy (85.00% and 90.00%) respectively.

Elevator Traffic Flow Prediction With Least Squares Support Vector Machines

Fuzzy Time Series Analysis Method Based on Least Squares Support Vector Machines

Traffic flow time series prediction based on LS-SVM

Traffic Flow Predicting Of Chaos Time Series Using Support Vector Learning Mechanism For Fuzzy Rule-Based Modeling

Passenger Flow Monitoring of Elevator Video Based on Computer Vision

Short-Term Forecasting of Traffic Flow and Speed: A Deep Learning Approach

Short-Term Traffic Flow Prediction Based on Least Square Support Vector Machine with Hybrid Optimization Algorithm

Short-term Traffic Flow Prediction Based on Optimised Support Vector Regression.

Peak Traffic Forecasting Using Nonparametric Approaches

Short-Term Traffic Flow Forecasting by Selecting Appropriate Predictions Based on Pattern Matching

Real-time Multistep Prediction of Public Parking Spaces Based on Fourier Transform–least Squares Support Vector Regression

Application on traffic flow prediction of machine learning in intelligent transportation

Short-Term Demand Prediction Method for Online Car-Hailing Services Based on a Least Squares Support Vector Machine.

Elevator traffic pattern fuzzy recognition based on least squares support vector machine

Dynamic least squares support vector machine

Prediction of Passenger Flow in Urban Rail Transit Based on Big Data Analysis and Deep Learning

Traffic flow forecasting based on phase space reconstruction and PSO-LSSVM

Fuzzy Hybrid Framework with Dynamic Weights for Short‐term Traffic Flow Prediction by Mining Spatio‐temporal Correlations

Incorporating CNN-LSTM and SVM with wavelet transform methods for tourist passenger flow prediction

Short-Term Load Forecasting Based On Support Vector Machine

Current-carrying Fault Prediction of Electric Equipment Based on Least Squares Support Vector Machine