T. Vasylets, O. Varfolomiyev, V. Ishchenko, S. Kovalchuk, O. Susla

Annotations languages:

** Description:** Performed synthesis and research of the operation performance indicators for the electromechanical system’s neural network control, accounting for the finite stiffness of the kinematic chain elements. Developed mathematical model of a two-mass control system for the generator-motor electric drive, which has a structure with a summing amplifier where reference signal and feedback signals a flexible current feedback and current feedback with cutoff. Two-mass system is modelled in MATLAB. It’s found that in the transient modes there are significant fluctuations in the affecting the distance accuracy of the specified displacements. of the neural network control to provide the target dynamic characteristi al network system is developed. For the neuro-control the NN Predicti Neural Network Toolbox is used to efficiently implement generalized predictive control using a multilayered feed-forward neural network as a control object model. To determine the performance indicators of the neural network control, the system was simulated using a scheme developed in MATLAB Simulink package. As a result of the simulation analysis it is established that the neural network system provides high control quality. Based on the obtained results it is concluded that the use of neural network technology in the design of modern control systems is a promising direction of research and requires further development.

**
Keywords:
** neural network technology, neural network control, two-mass electromechanical system, summation amplifier system, neural regulator with foresight NN Predictive Controller

Vasylets, T.Yu., Varfolomiiev, O.O., Ishchenko, V.S., Kovalchuk, S.L. and Susla, O.O. (2018), “Neiromerezheve upravlinnia elektromekhanichnoiu systemoiu z pruzhnymy zviazkamy v kinematychnykh peredachakh” [Neural network control of the electro-mechanical system with elastic constraints in kinematic transmissions],