SYNTHESIS OF THERMAL DIAGNOSTIC EXPERT COMPONENTS WITH AN ARTIFICIAL NEURON

Abstract

The article analyzes the automation tools in which artificial neural networks are used. It has been considered examples of effective use of hardware solutions with software versions of artificial neurons and other components, which allow to expand the functional properties of automation, while lowering the requirements for used computing facilities. On the example of electric motors intelligent technical diagnostics, it has been noted the possibility of assessing the technical state of complex electromechanical systems. The purpose of this work was to develop algorithms for computing and logical cycles suitable for the synthesis of a thermal diagnostic expert with an artificial neural network capable of identifying expected faults in electromechanical equipment of any complexity. It has been proposed a modular structure of the neural network software, an algorithm for the rate of temperature change, an artificial neuron module and other components. Simulation modeling and hardware implementation of an artificial neuron confirmed the suitability of the proposed solutions for the implementation of a diagnostic thermal examination device. The use of experimental data in electronic components made it possible to obtain a calibration characteristic for its subsequent use in assessing the development trend of possible non-standardized thermal events that appear when malfunctions are activated in certain parts of the operating equipment. It has been given variants of diagnostic expertise and identification of thermal events, preceding possible faults in the elements of electromechanical devices. It has been considered real examples of the main user program synthesis, taking into account access to the necessary settings of the hardware and control parts of the diagnostic device. It has been presented the illustrations of changing interfaces, which visual advantages lead to increased perception of the provided and associated computational functional properties of the diagnostic device when operating by users without special training. The applied solutions and visual examples of experimental and simulation modeling of the developed components of a thermal diagnostic expert with an artificial neural network are presented in the work.