Using fuzzy logic to diagnost the technical state of an object

The main task of diagnostics is to determine the technical state of the object in order to ensure its reliable and safe operation. It is assumed that there is an object, the technical state of which is characterized by a set of diagnostic parameters. From the values of these parameters is necessary to evaluate the state of the object: serviceable or unserviceable. This problem can be solved, for example, by methods of machine learning. However the problem consists in what can’t be defined in advance what of methods of machine learning will provide the correct solution of a task. In this paper, the use of fuzzy logic methods is proposed for diagnosing the technical state of an object. The rules of fuzzy logic make it possible to model the system in the case of the impossibility of applying traditional methods, and also, instead of exact mathematical calculations, it is more efficient to use qualitative assessments of the technical state of the object. However, fuzzy logic methods do not replace traditional approaches, but, on the contrary, complement them. To convert the clear input values of the diagnostic parameters of the object into fuzzy output, characterizing the technical state of the object, the Mamdani fuzzy inference algorithm is used. To assess the effectiveness of using fuzzy logic to diagnose the technical state of objects with two values of the output variable, the quality criteria of the binary classification are used: the F-measure and the AUC criterion. The effectiveness of the proposed approach is shown by the example of diagnosing the technical state of an object using eight specified parameters of its operation using fuzzy logic, as well as basic machine learning methods (logistic regression, discriminant analysis and a naive Bayes classifier). It is shown that the use of fuzzy logic can improve the accuracy of technical diagnosis by 5%-8% compared with the basic methods of machine learning.