Optimization of dynamic incentive strategies for public transportation based on reinforcement learning and network synergy effect

Abstract

Abstract Aiming at the challenges of peak passenger congestion, user behavior heterogeneity and insufficient network synergy faced by public transportation systems in urbanization, this study proposed the Dynamic Incentive Strategy-Heterogeneous Response Synergy Model (DIS-HARM). The model integrated reinforcement learning, user heterogeneity modeling and small-world network synergy mechanism, adjusted the carbon credit intensity in real time by dynamic incentive generator, quantified the diminishing marginal utility effect of incentives for high-income groups by combining elastic user identifiers, and designed weather attenuation coefficients to optimize the spread of social influence. Simulation results showed that DIS-HARM significantly improves system efficiency and fairness: the peak hour passenger flow reduction rate reaches 72.2% (2.5% higher than the static strategy), the average peak hourly cost is reduced by 3.125%, and 36.5% of the incentive resources are tilted to the low-income group (83.1% coverage rate) at the same time. The model provided a theoretical tool for dynamic pricing and differentiated incentive strategies for urban transportation management, helping to achieve the dual goals of green travel and social equity.