Sistema baseado em nuvem para o monitoramento de uma planta industrial

Abstract

Programmable Logic Controllers (PLCs) are central elements in industrial automation systems, with specialized control functions. Algorithms that involve complex mathematical operations are difficult to implement. However, its communication capabilities make it possible to make data available to cloud systems with a huge variety of tools for the development of monitoring systems based on machine learning. In this work, native PLC features were configured to send data to the cloud and receive the desired diagnostics. The cloud environment was developed for training models and monitoring an industrial plant, taking advantage of existing libraries. The result is a cloud environment with high computational power that provides important diagnostics to a SCADA (Supervisory Control and Data Acquisition) system in an industrial plant. Any browser screen on the local operating station allows parameterization of algorithms programmed in the cloud. In the developed application, algorithms for fault detection in the controlled plant were implemented in the cloud, using operating data for model training and monitoring. The results can be viewed both on the SCADA system graphical interface and on the cloud system dashboard. The proposed environment can be easily extended to industrial processes in general.