The Formation of Multi-Level Fuzzy Diagnosing Systems Structure Using Stochastic Models

Abstract

An appearance of diagnostic systems of complex structure dynamic objects, based on "soft computing" is connected, primarily, with a relatively large number of possible developing faults (abnormalities) in these objects. The number of symptoms for diagnosis is often limited, and the field of values of these characteristics for different pathologies are intersecting. The claimed genus can include objects of different nature: technical, chemical, biological, etc. A very common situation for such objects is the lack of sufficient statistical samples of values of signs of pathology for the successful application of statistical methods for diagnosis. This work discusses the problems of structural construction and functional multi-level diagnosing systems based on fuzzy logic reasoning. While not subject to the restrictions on the dimensionality of the system (the number of diagnostically pathologies) and the possible number of signs used for diagnosis. As a tool, providing an acceptable and functional structure of the diagnosing systems and its individual levels, is used stochastic model of the diagnosed object. The model implements the generation of random values of features with preset mathematical expectation and standard deviation values for the respective pathologies of the diagnostic object. This article paper proposes a method of constructing a multi-level diagnosing system in a sequence of steps (actions) to build appropriate levels of diagnostics. First of all it is solving the fundamental question of the expediency of constructing a multi-level (hierarchical) diagnosing system.