Somka O. Determination and forecasting of reliability of induction motors with long mean time between failures

The dissertation contains the analysis of a condition of induction motors and their structural units in the course of manufacturing, long operation and repairs. Its results reveal that most failures occur due to damage to the insulation of the windings and the bearing assembly, which is manifested in changes in temperature and vibration characteristics of the machine. The inapplicability in this case of existing approaches to determining and forecasting the reliability of the induction motor indicates the need to take into account both the interaction of major components and environmental conditions, which causes the need to improve existing models and develop new ones for calculating and forecasting reliability. The ways of conducting accelerated tests on the reliability of bearing assemblies of low- and medium-power induction motors are substantiated, which allows assessing the features of changes in their state at the last stage of the curve of their life cycle. The peculiarity of taking into account the obtained results when constructing mathematical models of bearing unit reliability is explained. It is shown that the use, even with a limited number of influencing factors, of mathematical models for calculating the reliability of structural units of induction motors, built on the theory of experimental planning, does not enable one to take into account changes in structural properties during production and long operation. It is determined that structures based on the joint application of the provisions of fuzzy logic and neural network theory are most effective for predicting the reliability of an induction motor. Taking into account the above, the structure of the diagnostic complex for testing of induction motors for reliability is substantiated and developed, which provides the necessary modes of operation and acceleration of their implementation due to the variation of temperature and vibration factors. The complex is designed as a unified device, test modes and use of which do not violate the physics of aging and wear of structural components and materials. Its main purpose is to obtain diagnostic and training information necessary for the construction of mathematical models based on fuzzy logic and neural network, and for predicting the reliability of an induction machine using developed models that take into account its actual state. For the case of low- and medium-power induction motors, the ratio for determining the reliability taking into account the real state of the magnetic system during long-term operation and repairs, which is characterized by the deterioration of its magnetic and electrical properties, is generalized. The possibility of improving the accuracy of determining the number of short-circuited turns in the windings of induction motors during repair and current operation based on the analysis of the nature of transients is found, and a mathematical model to determine reliability taking into account the state of the stator winding is obtained. As a result, an urgent research problem is solved, which is to improve and test models for determining and predicting the reliability of induction motors, which makes it possible to take into account the influence of informative state of structural units, their interaction with each other and basic environmental conditions.