A FUZZY KNOWLEDGE BASE OF BALL-BEARING WEAR AND ITS PRACTICAL APPLICATIONS

A fuzzy oriented expert system is used to integrate results of more than 2000 rather heterogeneous multi-dimensional (51 variables) ball bearing experiments. Only a subset of the variables is directly connected to wear processes. The expert system generates additional indirect information links between a subset of "wear" experiments via "non-wear" experiments. These links are information intensive and therefore very useful for wear studies. A fuzzy interpretation of experiments (fuzzy knowledge base) offers a simple and flexible formal tool for integrating inconsistent, sparse and vague knowledge items. A primary knowledge (experiments) is used directly by a modification of fuzzy reasoning algorithm. A man-machine dialogue (the fuzzy evaluation of the level of radial vibrations) is given in full detail).