State Observer based Elman Recurrent Neural Network for Electric Drive of Optical-Mechanical Complexes

This paper proposes an application of Elman recurrent neural networks as state observer to estimate electromechanical variable coordinates in electric drive control system of the optical-mechanical complex. The mathematical description of electric drive of the optical-mechanical complex is developed in the form of a two-mass elastic system. These elastic vibrations can be damped by using additional feedback signals from the elastic moment and the load velocity. The architecture of dynamic recurrent neural networks-based Elman scheme in investigated in the form of vector-matrix model, which allows approximating a wide class of nonlinear dynamic systems. During computer simulation in the MATLAB/Simulink environment, the comparison of the root-mean-square error between different learning algorithms for Elman's recurrent neural networks was carried out to study their accuracy estimates coordinates in a closed loop control system of optical-mechanical complex.