Submodel pattern extraction for simulink models

Abstract

Long before MDE became a popular method for software development, Simulink was firmly established as a tool for model--driven development of hybrid industrial systems. While practical and expressive for model creation and reuse, Simulink lacks for good abstraction mechanisms, and copy--paste--modify is a standard paradigm in Simulink development, leading to large numbers of variants of similar submodels. SIMONE is a framework and tool for automatically identifying, classifying and formalizing submodel patterns in Simulink models, using near-miss clone detection to find similarities and model merging to identify points of variance. The result is a set of submodel patterns which can be used as a reference for variance in the models, supporting consistency analysis, test optimization, fault identification and the disciplined generation of new subsystem instances using feature selection.