The maglev train achieves frictionless stable levitation. Nevertheless, track irregularities and high speed lead to fluctuations in the levitation gap. Thus, the control parameters of the levitation system must be adjusted to ensure safe operation. At present, the parameters’ adjustment mainly relies on expert experience and not adapted to dynamic changes. Therefore, this study proposes a reinforcement learning driven method for adjustment of the levitation system control parameters. Firstly, the levitation model considering speed and track irregularities is established. Secondly, a reinforcement learning driven control parameter adjustment method is presented. The control parameters are modified in real-time. Finally, simulation verification is conducted. Three typical speed scenarios are designed to test the levitation system over irregular tracks. The results indicate that after adjustment the levitation gap fluctuations are significantly reduced. Moreover, the control performance evaluation indicators also performed exceptionally well. The method is of great significance for ensuring the stable operation of maglev trains across the entire speed range.

Maglev trains, levitation system, reinforcement learning, parameters adjustment, track irregularities.