Telemach*: A Functional Approach to Face Problem Solving in Knowledge- Based Signal Processing Systems

Abstract

Abstract This paper presents TELEMACH, a problem solving system which provides syntactically uniform access to heterogeneously implemented knowledge bases. The system is aimed to use in autonomous signal pocessing ap plications such as e.g. , in the fields of knowledge-based system identification and fault detection. The applied knowledge representation scheme is functional in the sense that elemantary knowledge base constituents of the system are specified by their behaviour without any reference to how they are to be implemented. A functional interface is applied to the knowledge base which implements operations on the representation. On the platform of a procedural semantics we introduce the notion of the problem solving supervisor which, - in contrast with the conventional approaches, - exerts control of the reasoning process in a more flexible manner. As to the methodology, our recent approach formalizes the autonomous signal processing system as a discrete event dynamical phenomenon. In this theoretical framework a special action ontology, especially oriented to the signal processing domain, is presented which is based on temporal relations and the concurrent program viewpoint. Daedalus, the integrated graphical knowledge base editor of TELEMACH makes the development of complex signal processing strategies possible. An object oriented design philosophy is used to realize the idea.