Inductive acquisition of diagnostic knowledge for states tree with complex structure

The diagnostics of machinery is nowadays aided by expert systems. In order to carry out a diagnosis in such a way a knowledge base on the concerned machine is necessary. This knowledge may be acquired either from domain experts or from databases of examples. The paper deals with a method of diagnostic knowledge acquisition from examples. The method is also suitable in the case of complex structure of a set of technical states. The approach depends on optimal decomposition of the set of examples with respect to the criterion of the highest performance of the family of state classifiers, which are determined by the contents of the knowledge base induced from these examples. Moreover, some instances are presented that concern applications of the method for acquisition of diagnostic knowledge on technical states of rotating machinery. (C) 2001 Academic Press.