Many-objective optimization of multi-mode public transportation under carbon emission reduction

The incoordination between public transportation system construction and urban infrastructure development is a challenge for the sustainable development of cities. Exploring the associations between the built environment, transportation, and carbon emissions from a macro perspective is significant for realizing the goal of low-carbon urban transportation. This study proposes a collaborative optimization framework for the built environment and public transportation structure under carbon emission reduction. The purpose is to enhance the services capacity and emission reduction potential of public transportation with limited resources. Six factors are considered to construct the many-objective optimization model, including carbon emissions, energy consumption, government subsidies, time and economic costs, and road network resource occupation. The parameters of the model are calculated based on various multi-mode travel data (including vehicle order data, vehicle global positioning system (GPS) trajectory data, and intelligent card (IC) data) and built environment data. During the process, a data processing method for identifying bus drop-off points and inferring urban functional areas is developed. Then, the NSGA–III–DE (Differential Evolution) algorithm is designed to obtain the optimal solutions. The methodological framework is validated in the experiment implemented in Shenzhen city. According to the hypervolume (HV) value, the performance of NSGA–III–DE is compared with that of NSGA-II and NSGA-III. The results show that NSGA–III–DE has better global search ability and presents stable performance for different mutation operators. Finally, the optimization results are further discussed to provide effective guidance for urban planning and low-carbon transportation. •The associations of urban built environment, transportation and carbon emissions are explored.•A multi-mode public transportation optimization model under low-carbon target is proposed.•NSGA–III–DE performs a better global search ability in finding excellent solutions.•The data processing methods are provided for inferring bus drop-off points and functional area.