A Reinforcement Learning approach for bus network design and frequency setting optimisation

Abstract This paper proposes a new approach to solve the problem of bus network design and frequency setting (BNDFS). Transit must satisfy needs both service users transit operators. Numerous optimisation techniques have been proposed for BNDFS in literature. Previous approaches tend adopt sequential strategy that conducts routing two separate steps. To address limitation optimisation, our algorithm uses Reinforcement Learning simultaneous three key components BNDFS: number routes, route frequencies. The can best set routes without defining total advance, which reduce overall computational time. was tested on benchmark Mandl Swiss network. is further extended express services. validation includes additional test scenarios modify demand level be useful assist agencies planners improving existing cope with changing conditions.