Model Transformations for Knowledge Base Integration within the Framework of Structural Modelling

Abstract

A plethora of different methods and models has been worked out for model-based diagnosis of complex technical systems. Models are described using a distinctive language that is serious obstacle towards effective reuse of already developed models taken from different approaches. It remains valid even for rather similar modelling methods such as structural modelling and multilevel flow modelling. Both were invented several decades ago and have accumulated a wide spectrum of structural models used in different application areas. It is impossible to integrate these models in a knowledge base due to their syntactically different description languages. This paper reflects results obtained in a pilot project where transformation methods and algorithms have been worked out for structural models of complex technical systems developed in the structural modelling and in the multilevel flow modelling approaches. It is supposed that a set of knowledge bases constitutes a repository of syntactically different models developed for diagnosis problem solving. Problem of reuse and integration of models captured in a knowledge base is reduced to the problem of model transformations from one specific modelling method to another. Principles of transformation method are shown. Similarities of model elements are proposed. Two transformation algorithms are presented.