The setting you describe resembles non-stationary banditsnon-stationary bandits. In this setting, in which the means of the arms are allowed to vary over time.
The following papers might be of interest:
- A Linear Bandit for Seasonal Environments
- Stochastic Multi-Armed-Bandit Problem with Non-stationary Rewards
- On Slowly-varying Non-stationary Bandits
Here there are also have these lecture notes which seem relevant.