Application of neural network with adaptive interaction for speed control of the drive system with elastic joint

This article presents adaptive speed controller applied for electrical drive containing extended mechanical part of the structure. Load machine is connected with long, elastic shaft. Such construction leads to additional disturbation in control system. Oscillation of state variables are appearing. In described solution controller based on neural network is applied. Weights coefficients are calculated on-line during work of the drive. Adaptation of gains in neural network is realized according to adaptive interaction algorithm. Achieved results present correct work of drive in presence of load switching. Additionally control structure is tested against changes of the mechanical time constant of load machine. Simulations are verified by experiment realized on laboratory set-up.