METHODS OF MEASURING TORQUES OF ELECTRIC MOTORS USING ARTIFICIAL NEURAL NETWORKS

Abstract

Neural networks can be used not only in electric motor control systems to optimize their operation, but also in information and measurement systems that are used to control and diagnose the rotational parameters of electric machines. Such use will allow to reduce the errors that are associated with a number of destabilizing factors. This especially applies to the operation of measuring devices in conditions of uncertainty. Traditional diagnostic and control methods, such as harmonic analysis and spectral analysis, are often time-consuming. In this context, neural networks open new horizons, providing a fast and efficient alternative. In this regard, the article explores the possibilities of neural networks that can be used to analyze and identify patterns based on data obtained from voltage, current, angular velocity, angular acceleration, and torque sensors that are related to specific states and characteristics of rotary motion of electric motors. This ability of neural networks allows you to predict the torque of the electric motor, which can be used to adjust the output parameters of the measuring devices. As a result of the study, the method of using neural networks for evaluating the feedback signal from the torque sensor in the direct current drive was presented in order to determine the parameters of the torque. A threelayer neural network is proposed, which, after undergoing intensive training, was tested on a measuring bench in order to determine the optimal operating modes. Application of the model made it possible to increase the accuracy of torque measurement by predicting individual initial parameters that are affected by vibrations and temperature fluctuations that change the elasticity of dynamometric measuring elements.