Горный журнал УГГУ - Результаты поиска для: Babokin G. I

The structure is described of the organization and building of equipment for technical diagnostics of complex objects by the example of a mining electromechanical complex, which is based on a shearer-loader. Objects under control are the electromechanical systems (EMS) containing asynchronous engines of various power; hydraulic knots, the high-voltage and low-voltage switching equipment; electronic power converters (frequency, voltage, rectifiers); gearboxes of conveyor and mining shearer; transformers. Process of diagnosing is carried out taking into account factors of the external environment. The functional scheme of a diagnostic complex includes three levels of hierarchy. Each level of hierarchy acts as managing in relation to all subordinate and as operated, subordinated, in relation to the higher one. The lower level contains the sensors and transforming equipment measuring the parameters of EMS and factors of the external environment, modules of input-output and isolated barriers. The average level contains technical means of interfaces transformation and the subsequent collecting, temporary switching of telemetric messages. The top level includes set of the automated working places of a dispatcher, functioning under the control of special software. At the root of special software for the diagnosis of technical states are the neural network algorithms allowing to solve the problems of control and forecasting EMS technical states. These algorithms are opened and adjusted with a possibility of adding new diagnostic signs. The algorithms of diagnosing used in the program are based on the results of model and field observations and are object-oriented. It is shown that the developed equipment allows operating personnel to control technical condition of systems of any complexity on-line, including the electromechanical equipment for explosive atmospheres.

Main regulations and stages of neural network diagnostics of electromechanic systems of mining machines are represented. On the basis of diagnostic model lies multilayer neural network of direct expansion, taught by the method of indirect mistake expansion.