Model Based Diagnostics ? Today and Tomorrow

The paper presents author’s opinions concerning capabilities and limitations of the model based diagnostics. Present development in computer science and methodology of modelling has increased those capabilities considerably. It is obvious that talking about model based diagnostics assumes possessing not only an advanced theoretical model of the examined object but also models of irregular states and mutual relations between defects and their symptoms. Acquiring sufficiently reliable relations of defect-symptom type is a difficult, and frequently an extremely difficult task. The opinions presented in the paper concern one of the most intriguing phenomena, namely the formation of whirls and whips in slide bearings of a rotating machine. Although those phenomena are being the object of investigation in many research centers all over the world, their physics has not been satisfactorily recognized yet. The paper presents the abilities of computer simulation of the development of oil whirls and whips using the methods characteristic for model based diagnostics. Another very interesting issue is an assessment of the influence of a random character of certain input data ? in this case ? changes of external excitations of the system. This problem is related to the so-called heuristic models often placed in opposition to widely used algorithmic models. The main issue concerned the question whether the heuristic methodology can move to the techniques, in this case, the rotor dynamics. Heuristic approach may be the future of model based diagnostics. The objects of investigations were the high-speed rotor of a micro turbine being an element of the micro power plant in dispersed power engineering based on renewable energy sources. In both cases (whirls and heuristics), it is necessary to have very powerful tools for analyzing the behavior of objects, especially in the non-linear range. These tools are the basis of modern model based diagnostics.