To solve the problem of performance improvements automatic controller of cutting current of the shearer, the article analyzed the application of three possible options for regulators in the control system: the use of a PI-controller with constant coefficients, the use of an MPC-controller and the use of an adaptive PI-controller with a neural network corrector of its coefficients. The coefficients of a conventional PI-controller are often optimally selected for a particular state of the object and, when it transitions to other states, these coefficient values do not allow us to obtain transients required on quality. The neural network corrector was trained to change the PI-controller coefficients when changing such a parameter as coal strength. As a result of modeling in MATLAB / Simulink, it was confirmed that the proposed control scheme is more efficient than PI-and MPC- control.
Paper
Full text
Neural Network as a Corrector of Performance of PI-Controller in Regulator System of Shearer
Semantic Scholar · Engineering · 2020
Abstract
To solve the problem of performance improvements automatic controller of cutting current of the shearer, the article analyzed the application of three possible options for regulators in the control system: the use of a PI-controller with constant coefficients, the use of an MPC-controller and the use of an adaptive PI-controller with a neural network corrector of its coefficients. The coefficients of a conventional PI-controller are often optimally selected for a particular state of the object and, when it transitions to other states, these coefficient values do not allow us to obtain transients required on quality. The neural network corrector was trained to change the PI-controller coefficients when changing such a parameter as coal strength. As a result of modeling in MATLAB / Simulink, it was confirmed that the proposed control scheme is more efficient than PI-and MPC- control.