Skip navigation
Please use this identifier to cite or link to this item: https://libeldoc.bsuir.by/handle/123456789/38689
Title: Adaptive control of robotic production systems
Other Titles: Управление технологическим циклом производства на основе модели нейроконтроллера
Authors: Smorodin, V. S.
Prokhorenko, V. A.
Смородин, В. С.
Прохоренко, В. А.
Keywords: публикации ученых
technological production process
parameters of operation
probabilistic network chart
state indicators
methods of adaptive control
neural network
Issue Date: 2020
Publisher: БГУИР, РБ
Citation: Smorodin, V. S. Adaptive control of robotic production systems/ Alena V. Shviatsova, Vladislav A. Prokhorenko // Открытые семантические технологии проектирования интеллектуальных систем = Open Semantic Technologies for Intelligent Systems (OSTIS-2020) : сборник научных трудов / Белорусский государственный университет информатики и радиоэлектроники ; редкол. : В. В. Голенков (гл. ред.) [и др.]. – Минск, 2020. – Вып. 4. – С. 161-166.
Abstract: The purpose of the work, that is presented in this paper, is to develop a method for adaptive control of a technological production cycle based on a software and hardware system that includes indicators of the hardware units states, parameters of the technological production cycle operation, simulation model of the probabilistic technological process and a built-in decision-making system. Operational interaction of the software and hardware system components and construction of the feedback control connections is implemented through the control parameters and variables of the simulation model based on the output of the neuroregulator model. To address the described problem, tasks related to implementation of the neural network technologies when constructing architecture and mathematical model of the neuroregulator were solved. The mathematical model of the neuroregulator is based on parameters of operation of the physical prototype, construction of the feedback connections for the real-time control (adaptive control) is based on the procedure of training of a recurrent neural network that includes LSTMcells. Considering the testing results is was found out that recurrent neural networks with LSTM-cells can be successfully used as an approximator of the Q-function for the given problem in the conditions when the observable region of the system states has complex structure.
URI: https://libeldoc.bsuir.by/handle/123456789/38689
Appears in Collections:OSTIS-2020

Files in This Item:
File Description SizeFormat 
Smorodin_Adaptive.pdf182,82 kBAdobe PDFView/Open
Show full item record


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.