Sistema especialista para ambiente industrial baseado em regras com auto-aprendizagem

Abstract

This work presents a methodology for knowledge acquisition and representation through automatic logic rules for an industrial plant. Initial knowledge of an industrial process can be gained through a specialist who interprets situations present in the plant and can describe what is happening. In this paper, we present a way to acquire statistical knowledge of the plant during the execution of its processes, using an online clustering method known as TEDA-Cloud, modified for performance increase. Knowledge representation is described through the manipulation of a neural network known as CILP (Connectionist Inductive Learning and Logic Programming) and a proper symbology is described to represent the logical variables taken from the process signals. The results show an efficiency in interpreting the rules and acceleration in the clustering process and classification of the standards that define the rules.