This work develops a methodology for the optimal management of individual manufacturing processes within a process chain. To control a process based on its inaccessible state, real-time capable and robust observer models are created using statistical modeling to reconstruct the process state from a few observable measurements. State tracking and optimal process management are implemented with a generic process model for the general description of process relationships, which are shaped by adapting to specific data when applied to a process. The feasibility of generic process modeling is demonstrated through its application to several deep drawing processes of varying complexity and a resistance spot welding process.