A Subsystem Division Method by Clustering

State explosion problem is the primary obstacle to model complex system with Petri nets; modularization and hierarchy provide ways to solve this problem. When the bottom-up method is adopted, system functions in the lower layers are combined to obtain sub systems. The idea of clustering is introduced to decide which functions should be combined. The operation to combine two functions is defined; the distance between two functions is calculated by the degree of relevancy; a clustering algorithm with the idea of nearest-neighbor-first is designed to divide functions into sub systems. The function clustering method presented in this paper facilitates the Petri nets based system developing process.