WU Peng, YU Zewei, CHU Chengbin
In recent years, the emergence of customized buses has provided convenient services for commuters. However, changes in road conditions, attributed to factors such as traffic congestion, bad weather, and car accidents, have increased passenger travel time and significantly impacted the punctuality of bus operations, diminishing the appeal of customized bus services. This study addresses a new commuting customized bus network design problem. It takes into account uncertain road conditions, passenger separation, and the presence of heterogeneous vehicle models. We first formulate it into a multi-objective robust optimization model based on the minimum-maximum regret criterion. This model optimizes the selection of heterogeneous vehicle models, fleet routes, travel times, and passenger assignments with the aim of minimizing both passenger travel costs and bus system operating costs. Then, to effectively solve the model, we propose a multi-objective hybrid adaptive large neighborhood search algorithm. This algorithm incorporates problem-specific characteristics, introduces a passenger separation destruction operator based on these characteristics, and integrates a crossover operator to enhance the algorithm's optimization capability. Results of extensive numerical experimental results indicate that: i) The consideration of passenger separation and heterogeneous vehicles leads to improved vehicle resource utilization efficiency. Considering passenger separation results in an average time savings of 2.44% and a reduction of 11.46% in operating costs. Similarly, the inclusion of heterogeneous vehicle models yields an average time savings of 1.74% and a decrease in operating cost by 24.95%. ii) Compared with the traditional large neighborhood search algorithm and NSGA-II, our proposed algorithm consistently produces higher-quality Pareto solutions. iii) when compared to solutions obtained under deterministic road conditions, the solutions obtained in the presence of uncertainty in road conditions achieve an average reduction of 1.42% in passenger travel cost and a decrease of 18.24% in operating cost. This confirms the necessity of conducting robust commuting customized bus network design research under uncertain road conditions.
