SYNTHESIS OF NARMA-L2 NEURAL CONTROLLER FOR TRACTION INDUCTION ELECTRIC DRIVE SYSTEM

Abstract

In today's world, where technology is developing rapidly, electric drive systems play a key role in a huge range of applications: from industrial production lines to vehicles and household appliances. These systems require high efficiency, accuracy and reliability. It is here that there is a need to improve and optimize the processes of controlling electric drives.In recent decades, neural network controllers, or neuroregulators, have won their place in the world of automatic control. Their unique ability to model complex nonlinear dependencies makes them an ideal tool for applications in electric drive systems. Neuro-regulators can adapt to changing operating conditions, learn from data and optimize the control process depending on specific requirements and conditions.This work is devoted to the possibility of using a neuroregulator in the traction asynchronous electric drive system. In the Matlab/Simulink environment, a simulation model of the AD914-U traction asynchronous electric motor vector control system described in the d,q rotating coordinate system was developed. The synthesis of the NARMA-L2 neuroregulator was performed, which combines the principles of the autoregressive model and the moving average model to provide prediction and control of complex processes. The main idea of this controller is to build a nonlinear transformation of input data that can predict the future states of the system. To demonstrate the capabilities of neuroregulators in traction electric motor control systems, comparative modeling of the NARMA-L2 neuroregulator and the classical proportional-integral regulator was conducted.The results of simulation modeling show that the system with a neuro-regulator shows the best indicators of regulation of the given parameters in transient processes and is a promising tool in the development of high-performance and energy-efficient traction electric drives.