Balancing electric bus operational performance and dynamic wireless charging benefits: A simulation-based bi-level dedicated bus lane deployment approach under uncertainty

Dedicated bus lanes (DBLs) are becoming increasingly important in electric bus (EB) systems because they can simultaneously provide transit priority and serve as favorable corridors for dynamic wireless charging. However, these two roles may spatially compete in DBL deployment. To address this issue, this study develops a simulation-based bi-level optimization framework for DBL deployment under uncertain EB travel times. The upper-level model seeks to identify a DBL deployment scheme balancing EB operational performance and charging benefits, while the lower-level model simulates the actual EB operations under a given scheme and provides simulation-based performance evaluations for the upper-level model. The time-consuming simulation and the uncertain EB travel times give rise to a stochastic and computationally expensive optimization problem with black-box evaluations. A novel topology-aware Bayesian optimization algorithm is proposed to solve this challenge. Numerical experiments are conducted based on a network with six bidirectional lines in Changsha, China. The embeddings processed by the algorithm increase the correlation coefficient between scheme similarity and performance similarity from 0.09 to 0.76, and the identified important segments become increasingly different as the differences in objective values increase, demonstrating the algorithm's ability to capture both global and local similarities among schemes. Ablation experiments further demonstrate that the proposed algorithm can efficiently identify DBL deployment schemes with lower objective values under a limited evaluation budget. In addition, comparisons between cases with and without the lower-level model reveal deviations of up to 268.25 s in bus headways and 1690 Wh in charged quantity, validating the necessity of this model for scheme evaluation. Moreover, compared with the other three schemes in terms of passenger travel time, passenger waiting time, EB headway, and charged quantity, the optimized scheme provides a well-balanced trade-off between EB operational performance and charging benefits.