Method and device for controlling a process within a system, in particular a grinding process in a grinding device

A method for controlling a process within a system, in particular a grinding process in a grinding device, comprising:

• detecting state variables (st) of the system with the grinding device, wherein similar state variables (st) of the system are grouped together and wherein the classification of similar state variables (st) of the system into groups is self-learning and self-optimizing and compares the costs of grouping state variables (st) of the system into a group with the costs of creating a new group;
• creating at least two process models (PM) from each group of state variables (st) after each group of state variables (st) is created, which each describe predicted effects of actuating actions (at) on the group of state variables (st) of the system, a structure of the at least two process models (PM) differing from one another, wherein at least one of the at least two process models (PM) is created for short-term effects of the actuating actions (at) on the group of the state variables (st) of the system and at least another one of the at least two process models (PM) is created for long-term effects of the actuating actions (at) on the group of the state variables (st) of the system;
• controlling of the grinding process within the system by executing actuating actions (at) under consideration of predefined control objectives and a process model of the at least two process models (PM) which currently provides the best prediction for the grinding process within in the system;
• and
• wherein grouping similar state variables (st) of the system into groups reduces the process models (PM) of the state variables (st) to save resources.