Vehicle maintenance constitutes a critical area of activity, with a major impact on both operating costs and service quality. However, relatively few research papers have been published on models and algorithms that could help optimize the maintenance plan for public transport. This extended abstract describes the main components of a new model and algorithm that have been successfully implemented for this purpose in the city of Angers in France. The paper was presented at the 2018 Conference on Advanced Systems in Public Transport, CASPT2018.