To enhance the accuracy and timeliness of urban transport carbon emission assessment under the “dual carbon” goals, the trends and driving mechanisms of residents’ travel-related carbon emissions were analyzed from both historical validation and future projection perspectives. An agent-based model was employed to simulate changes in residents’ travel mode choices under policy interventions, and the resulting public transport share was incorporated into a system dynamics model to construct a comprehensive prediction framework consisting of five subsystems, namely population, economy, transportation, green space, and residents’ travel carbon emissions. The proposed model was validated using historical data from Fuzhou and was applied to simulate residents’ travel carbon emission trajectories under multiple policy scenarios. It is found that private cars will remain the dominant contributor to residents’ travel carbon emissions before 2030. Although a modal shift from private transport to public transport can significantly reduce carbon emissions, carbon peaking cannot be achieved under single-policy scenarios. A substantial carbon peak before 2030 can only be realized through the coordinated implementation of multiple policies, with a maximum emission reduction rate of 29.50%. It is demonstrated that the integration of individual behavioral responses with system-level dynamics provides an effective approach for evaluating urban transport carbon reduction pathways and offers decision-making support for low-carbon transport policy formulation.