ZANELLA, Tyene Zoraski 1; MEZA, Gilberto Reynoso2;


Introdução:PID controllers are the main types of controllers applied in industrial processes. Its performance has a great impact on the sustainability, profits, and safety of operations in the industry, being a recurring subject in research and studies. With the arrival of industry 4.0, systems tend to become more complex and as a result, the number of control loops increases, reaching hundreds running simultaneously. This makes it even more difficult to ensure its smooth operation, opening the opportunity for the use of optimization algorithms and machine learning.

Objetivo:The aim of this project was to develop a prototype of a portable learner/optimizer device for data analytics of PID control loops.

Metodologia:The approach of this study was to conduct a set of steps regarding self- study on the related topics as machine learning, optimization, and control performance; hardware definition, how the prototype should be created based on the objectives of the study, selecting a microcomputer, a Raspberry Pi; communication architecture, selected based on internet of things (IoT) concept, which allows different software and hardware to communicate and exchange data; and validation the use of a benchmark to create a dataset as close to a real case as possible.

Resultados:The project architecture consisted of a microcomputer responsible for running a server in Node.js and optimization algorithms developed in Python. When this microcomputer is connected to the industrial process to be monitored, it can be controlled through a developed web application, where the process data can be uploaded and viewed, as well as results obtained by using classification algorithms.

Conclusões:The prototype developed in this study from the tests performed presented good functionality in all tests and communication protocols. The web application developed is user-friendly, responsive and simple, as well as the graphical representation of the results and the availability of various classification methods.

Palavras-chave:PID. Control performance. Learner. Web application.


    1. Estudante
    2. Orientador