In a paper published in the journal Scientific Reports, researchers proposed a novel three-tier artificial intelligence (AI) --based scheme for allocating carbon-neutral mobility hubs. The first tier identified optimal sites using a genetic algorithm (GA). In contrast, the second tier involved an ensemble-based suitability analysis of the identified locations, incorporating factors such as land use mix, population and employment densities, and proximities to parking, biking, and transit.
GA selected sites along with suitability rankings and scores. Image Credit: https://www.nature.com/articles/s41598-024-62701-z
The third tier employed a traffic assignment model to evaluate the sites' environmental and economic impacts. The identified hubs reduced daily vehicle travel and led to significant cost savings. The comprehensive approach integrated carbon-focused analyses and post-assessment evaluations for sustainable mobility hub planning.
Related Work
Previous works have shown that sustainable mobility involves a comprehensive shift in transportation planning, encompassing environmental, economic, and social equity. Transit-oriented development (TOD) and the '15-min city' model are essential concepts in this shift, aiming to create environmentally friendly and socially vibrant urban spaces.
Mobility hubs are innovative solutions that can significantly reduce private vehicle dependence and promote sustainable modes of transportation. However, existing research on mobility hubs has limitations, such as a narrow focus, over-reliance on heuristic methods, bias in decision-making, and lack of post-assessment evaluation.
Mobility Hub Selection
The proposed mobility hub location selection methodology consists of a hierarchical three-stage approach. The first stage involves travel time-based optimization using GAs to efficiently search for extensive solution spaces and select an optimal set of hubs.
The second stage deploys a sophisticated set of evaluative criteria that combine traditional indicators with the innovative 'parking proximity' element for suitability analysis. Ensemble methods such as Random Forest Regressor, XGBoost, and Gradient Boosting are used to weight factors based on their impact on carbon emissions.
In the second stage, the study transforms raster data, such as land use, to a vector format and calculates the ensemble-based weights for factors, enabling district-level aggregation and normalization. The 'Business Analysis' feature of the Arc geographic information system software suite (ArcGIS) online is utilized for its accuracy and flexibility.
A key innovation of this research is the inclusion of parking proximity and the application of ensemble methods to weigh various factors concerning their contributions to carbon emissions, reducing bias in the selection process and addressing environmental concerns.
The third and final stage focuses on implementing a quantitative post-assessment that uses traffic assignment to gauge the selected locations' benefits in terms of travel time savings and reductions in carbon emissions. The study leverages traffic assignment modeling to evaluate the chosen sites before and after implementing mobility hubs. Employing the Emme software, this approach not only facilitates the validation of site selection for mobility hubs against pragmatic goals but also demonstrates their contribution to altering modal shares and reducing carbon emissions.
The results of the quantitative post-assessment indicate notable economic benefits, including reductions in operational costs, travel time, traffic accidents, air pollution, and energy consumption. The analysts quantified these benefits following the guidelines of the Korean preliminary feasibility study, which offers a comprehensive framework for evaluating the positive impacts of mobility hubs.
Sustainable Mobility Hubs
The study optimized the selection of mobility hub locations using a GA, which emphasizes the survival of the fittest through crossover mutations. This method systematically improves the choice of locations and significantly increases the likelihood of identifying optimal hub sites.
In the second stage of hub location selection, the study integrated ensemble methods to transcend the constraints of traditional multi-criteria decision-making (MCDM) tools, enhancing accuracy and promoting sustainability in the decision-making process.
The carbon emissions factor showed a lognormal distribution, with most values centralized around the mean. The analysis indicated a positive correlation of 0.49 between the carbon emissions factor and parking accessibility and a negative correlation of − 0.43 with public transit and cycling accessibility.
The study employed ensemble methods to identify key variables for accurate predictions, with the random forest (RF), gradient boosting, bagging, and XGBoost regressors showing exceptional performance. The importance scores of the factors extracted from the best-performing ensemble algorithms revealed that cycling, parking, and public transit had the highest contributions to carbon emissions.
In the selection phase, the previously calculated weight values were crucial for conducting a suitability analysis and identifying the most sustainable hub locations. The five districts with the highest suitability scores were Yeoksam 1-dong, Myeongdong, Yeouido, Jongno 1, 2, 3, and 4-ga-dong, and Hoehyeon-dong.
However, the study refined the selection by comparing suitability scores across different regions, leading to the final selection of hub sites. The strategic integration of mobility hubs into these districts promises to enhance accessibility and support each area's distinct economic, cultural, and educational fabric.
Conclusion
This study presented a three-tiered approach for mobility hub location selection, addressing previous research gaps by integrating various factors and combining heuristic GA and ensemble methods. The results showed significant travel time savings, carbon emission reductions, and economic benefits, estimated at 225.5 million USD annually.
The study's framework aligns with sustainable development goals (SDGs), TOD, and the '15-min city' concept, but its findings are limited to Seoul. Future research should consider a broader range of factors and involve the community in decision-making.
Predicting Jack Speed and Torque of a Tunnel Boring Machine Using Artificial Intelligence