A FUZZY INTERPOLATION OF MULTIDIMENSIONAL EXPERIMENTAL RESULTS OF 80 NI 20 CR ALLOY

Relations among different parameters of alloys are very complex, integ rated, difficult to measure and therefore relatively ill-known. A fuzz y knowledge base is a flexible framework for acquisition of vague, spa rse and inconsistent knowledge. The fuzzy reasoning is to some extent equivalent to a conventional interpolation algorithm capable of interp olating among variously accurate non-equidistant points in a multidime nsional space. The pooling of the experimental records represents a tr ade off between minimal modification of the original data and eliminat ion of obvious inconsistencies among all sets of data. This is an ad h oc and time consuming process that requires highly qualified field exp erts and knowledgeable engineers. A sequence of several modifications is needed to decrease the expected inconsistencies. A general theoreti cal background for the methodology of pooling does not exist. Therefor e the paper presents a detailed description of a subset of data, based on a relatively homogeneous set of measurements (six variables and 12 0 fuzzy conditional statements). No a priori knowledge of fuzzy mathem atics is assumed. (C) 1997 Elsevier Science Ltd.