extracting, by one or more processing units at a current timeframe, an aging model from a control system managed by a machine learning agent, the aging model defining one or more predicted operating conditions of the control system for a future timeframe;

deriving an aging term from the aging model for quantifying the one or more predicted operating conditions for the future timeframe;

extracting, by the machine learning agent, a plurality of states from the control system defining a plurality of associated operating conditions of the control system;

receiving a goal statement from an administrative entity, the goal statement comprising one or more objectives for the control system in a natural language format;

parsing the goal statement for the one or more objectives to construct an optimality function comprising one or more terms representing the one or more objectives and the aging term;

determining one or more actions based on the plurality of states and the aging term for application to the control system at the current timeframe to modify the one or more operating conditions of the control system;

extracting a changed plurality of states and an updated aging term from the control system in response to applying the one or more actions to the control system;

calculating, by the optimality function, a measure of optimality for the one or more actions based on the changed plurality of states and the updated aging term; and

determining of one or more updated actions based on the changed plurality of states and the updated aging term for application to the control system at the future timeframe to modify the one or more operating conditions of the control system and increase the measure of optimality.