Title: Adaptive PID Controllers Tuning: LMS Gain Scheduling Training And Industrial Programmable Logic Controllers
Authors: Nunes, Victor Pereira; Farias, Hugo Baluz Bezerra de; Fonseca Neto, João V. da; Costa Filho, Pedro Turibe
Resumo: The main issue of this article is the blending of LMS training with the artificial neural network theory for design and improvement of PID type controllers tuning. The LMS method and the neural theory contribute for the gain adjustments and to establish the plant-activation function, respectively. Thus, the PlantPID dynamic system is seen as a single entity, providing the base of the proposed method for controllers design through ANNs. The proposed method is based on a stochastic optimization structure which puts together the target variables and its restrictions. This structure is solved according to an extension of the LMS method, taking into account the adapting controller design of programmed type. These models form the base for the development of logarithms to be implemented in the programmable logic controllers. The PID adaptive control system is synthesized in industrial programmable logic controllers due to its high computing power. The gain adjustments via LMS training is used to evaluate the DC motor speed control performance of the method.
Keywords: Artificial Neural Network; PID tuning; Industrial Process; PLC; LMS and DC electrical motor
Paper as PDF: 50100064.pdf
BibTex file: 50100064.bib