A Unified Demand-Diversion Prediction Approach to Real-Time Group Route Guidance

Abstract

Dynamic origin-destination (OD) demand estimation and prediction (DODE) and drivers' route diversion behavior (DRDB) are two fundamental determinants in the development of effective group route guidance (GRG) strategies for balancing traffic loads and enhancing system performance. This work presents a unified approach to incorporate DRDB into a DODE model. This model accounts for the influence of route diversion on DODE and predicts the OD demand in information provision environment. This study also develops a dynamic aggregate DRDB model based on the traffic data obtained through online detection and integrates this model with the DODE model. The test results of the case study show that the maximum deviation between the real OD volumes and the proposed DODE model under information provision, mean absolute percentage error, and normalized root mean square error is approximately 11.46%, 4.53%, and 5.29%, respectively. The integrated demand-diversion prediction model can accurately estimate and predict possible DRDB and the effect of traffic information on OD demand prediction using real-time traffic detected data. The model can also enhance the accuracy of OD demand and traffic state prediction under information provision, consequently increasing the effectiveness of the proposed GRG strategies.