PSI - Issue 20

E.F. Dubinin et al. / Procedia Structural Integrity 20 (2019) 103–107 E.F. Dubinin and V.I. Kuksova / Structural Integrity Procedia 00 (2019) 000 – 000

106

4

noncriticality to lack of a part of the source data in the process of logical inference.

Fig. 1. Graph of the membership functions of the variable «Reliability of diagnosis»

Several options of formation of rules for fuzzy inference and of the resulting model are possible:

2.2.3.1. Fuzzy inference rules providing creation of model without the generalizing parameter

Let some state of the diagnostic object Y i , ( i=1,n ) be characterized by a set of values of the used diagnostic parameters X j , ( j=1,m ): Y i =F ( X 1 , X 2 , …, X j ,…, X m ). Fuzzy inference rules should provide the following: a) to lead all possible states of OD to two types – workable and inoperable (serviceable and faulty) In this case it is necessary to provide splitting of the set Y into two subsets – workable Y 1 and inoperable Y 2 states. At the same time it is necessary by Mashoshin (2007): to formulate rules of splitting a set Y into two subsets – workable Y 1 and inoperable Y 2 states; to define criterion for assessment of degree of operability of an object and its belonging to one of classes in a subset Y 1 ; to establish signs of the arisen failures (to distinguish between states in a subset Y 2 ). b) to lead all possible states of OD to several, for example to stable, rather stable, unstable, emergency In this case it is necessary to provide splitting of the set Y into four subsets – stable Y 1 , rather stable Y 2 , unstable Y 3 and emergency Y 4 states. At the same time it is necessary: to formulate rules of splitting a set Y into four subsets – Y 1 , Y 2 , Y 3 , Y 4 states; to define criteria for assessment of stability of work of an object and its belonging to a subset Y 1 ; to establish signs of the arisen failures (to distinguish between states in subsets Y 2 , Y 3 , Y 4 ). 2.2.3.2. Fuzzy inference rules providing creation of model with the generalizing (defining) parameter This parameter can be defined as follows: a) from private informative parameters generalizing parameter (а criterial parameter), the change of which maximally characterizes the quality of OD, is selected In this case, the generalizing parameter is critical in relation to the change of private parameters and characterizes the onset of the critical state of the object.