TY - JOUR AU - Orlovskyi, I. A. AU - Gorobec, E. I. PY - 2014/01/13 Y2 - 2024/03/29 TI - USE OF NEURAL CONTROLLER IN ELECTROMECHANICAL SYSTEMS JF - Radio Electronics, Computer Science, Control JA - RIC VL - 0 IS - 1 SE - Control in technical systems DO - 10.15588/1607-3274-2014-1-26 UR - http://ric.zntu.edu.ua/article/view/27328 SP - AB - The paper deals with the relevance of improved synthesis methods for control systems of nonlinear objects using neural networks, as those allowing to substantially remove the mathematical problems of analytical synthesis and the designed system analysis. Besides, the review of the articles and mathematical description of three neural controllers (NC) are presented: predictive control based on the model of nonlinear autoregressive moving average and NC with the reference model. NC research tools of MATLAB are described. Current control and speed control loops for one-component and two-component electromechanical systems with DC electric drive are conducted. Mathematical modeling showed the possibility of high precision control of NC based on model of nonlinear autoregressive with moving average predictor and predictive control. NC with reference model needed too much training time, the quality of the transition process was low. ER -