A fleet of systems that is used to meet a demand for production that varies over time is considered. For example, a wind farm consisting of N wind turbines under contract with a customer must produce and supply a given power to the grid. To guarantee the power delivery availability, the wind farm is over-dimensioned with respect to the contractually binding power production. However, when producing the deterioration phenomenon worsens with the production level, leading an increased deterioration rate. Hence this study proposes a load-sharing decision rule based on prognostic information for a wind farm. The load-sharing decision rule redistributes the production objective on each wind turbine through the so-called control level according to its degradation state and corresponding expected production at each time when the prognostic information is available. The optimal control level of each wind turbine at each decision time is optimized according to both the production objective and the expected degradation. A global model is developed including all these features and implemented to solve the joint production/maintenance optimization problem by simulation. Simulations results show how the joint load sharing/maintenance approach allows to reach an optimum operation at the farm level.