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

105

3

2.1.2. Allocation from the created set of parameters of the subset having essential value for accident-free (trouble free) functioning of OD The choice of a set of diagnostic parameters (DP) for solving diagnostic problems is a multi-alternative task and is determined by such factors as the objective function of OD; maintenance strategy; used tools and technical diagnostics methods; time to diagnose; cost of diagnostic tools and the diagnostic process itself. The chosen DPs have to be informative so that the process of recognizing the state of the OD can be performed. The syntactic information content of a parameter, determined by the amount of obtained information without regard to its substantial value, can be defined relatively easily. Assessment of semantic and pragmatic value of this or that DP is poorly formalizable task. One of the possible ways to solve it is to use the apparatus of the theory of fuzzy sets. 2.1.3. Creation of information model of an object of diagnostics which describes its behavior and modes of functioning taking into account the parameters chosen at the previous stage The task of STD is exact diagnosis based on the results of DP analysis. The limitations may be the requirements of minimization of control operations; such indicators and characteristics as duration, reliability, completeness of control of technical condition of the object of diagnosis, depth of search of the place of refusal (malfunction), conditional probabilities of undetected and false refusal (malfunction), quantitatively defined for the technical diagnosing of considered object. 2.2.1. Definition of factor space of the fuzzy model. Reduction of all chosen parameters to fuzziness conditions Main classification signs of ways of formalization of fuzziness: by type of representation of a fuzzy evaluation of variable (of a fuzzy set); by type of the domain of values of the membership function; by type of the domain of definition of membership functions; by type of correspondence between the domain of definition and the domain of values; by the sign of homogeneity of the domain of values of the membership function; proved by Ibragimov (2009), Akhmetkhanov et al. (2015), Sibikina (2016). The selection of membership functions and the formation of rules of fuzzy output are, to a certain extent, subjective procedures, since, as a rule, they are determined expertly, heuristically (paragraph 2.3.). Fig. 1 shows an example of the membership functions of the variable «Reliability of diagnosis» (measured by the probability of obtaining a correct diagnosis of the state of the object being diagnosed). 2.2.3. Formation of rules of fuzzy output The most common model of knowledge representation in fuzzy models and expert systems built on their basis are production rules. The main requirements for the type of the production rule and algorithm of a logical inference for solving the problem of diagnosing the state of a complex object are by Katasev (2013): possibility of using different types of input and output parameters; possibility of processing of clear and fuzzy input data; consideration of significance (weights) of conditions in the rule; consideration of significance (reliability) of each rule; 2.2. Creation on the basis of the information model OD of a fuzzy model describing his behavior 2.2.2. Selection of membership functions