Abstract
The article considers the question of replacing the classic PI controller in the shearer control system with an adaptive device, in which the regulator coefficients are adjusted depending on the random change in the re-sistance of coal to cutting. The classic PI controller in the control system has constant proportional and inte-gral coefficients. However, despite the ease of setup and practical implementation, as well as the relatively high robustness, this class of control devices cannot ensure the optimal functioning of the control system in all modes due to the nonlinearity of the control object and the random nature of the coal strength changing as the shearer moves in the coal face. To overcome these shortcomings, a neural network implementation of tuning the coefficients of the PI controller is proposed. The possibility of correcting the coefficients of the PI controller controlling the speed of movement of the shearer, with a random nature of changing the strength of coal, is proposed and experimentally proved. It is shown that the coefficients of the regulator vary according to a ra-ther complex law. It is proposed to use a neural network of the multilayer perceptron type as a corrector of the PI controller coefficients. Neural network training was carried out by the Levenberg-Marquardt method. The correctness of the results was confirmed by the results of computer modeling. It is shown that the use of a PI controller with a corrector in the form of a neural network in the control system will increase performance of the load regulator by an average of 1.5–3 times in comparison with the classical regulator. All this will allow to avoid critical overloads, and hence the possible breakdown of the mechanical parts in the transmission of the shearer in case of a sudden collision of the working body of the shearer with a solid inclusion. The pro-posed adaptive PI controller can be further used to improve the control system of the shearer, which is able to function effectively in various modes.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have