Automatic Partitioning of Problems through Submodel Decomposition— a Promising Technique of Digital Enterprise Technology

This paper presents two main groups of results in the field of process modeling; first, highlighting complexity-related properties shared by several, if not all, levels of production; second, a family of methods set up to handle the aforementioned problems. The presented algorithms lead up to a submodel decomposition method combining generalized feature selection and artificial neural networks which further research may extend to an automatic determination of agent boundaries in complex systems. Aside from theoretical presentation, practical results in various levels of actual industrial production demonstrate the feasibility of the methods, suggesting that they can cope with complexity-related problems of different production levels in a uniform way.