METHOD FOR DIAGNOSTICS OF ASYNCHRONOUS ELECTRIC MOTOR BASED ON NEURAL NETWORK ANALYSIS Russian patent published in 2024 - IPC G01R31/34 

FIELD: electrical engineering.

SUBSTANCE: invention relates to electrical engineering, namely to means of increasing reliability of electrical equipment of critical technological cycles of enterprises and diagnosing the state of insulation of stator windings of asynchronous electric motors. Method includes recording of dependences of voltage and current consumed by an electric motor, followed by passing the obtained signals through a software filter of high and low frequencies and processing the obtained data using software, at the output of which a quantitative and qualitative assessment of the technical state is obtained. At the same time instantaneous values of currents and voltages are measured by synchronizing signal in distribution device or power panel feeding electric motors. Instantaneous voltage values are measured by a voltage sensor common for all electric motors. Instantaneous current values are measured by current sensors of individual microprocessor devices for electric motor protection transmitting the measured values to the data collection point. Data processing is performed by means of individual neural networks. Separate types of faults are detected. After shutdown and termination of motor run-out, test voltage is supplied, and active component of current is calculated from current value through insulation and its phase shift, and residual life of stator insulation is estimated.

EFFECT: possibility of diagnosing developing defects of insulation at early stage of development and estimation of residual life of electric motor insulation.

1 cl, 3 dwg