Intelligent Controller for Industrial Processes Applied to a Distributed Two-Tank System

Abstract

This paper proposes a self-evolving intelligent controller, tested under a proof of concept of a purely virtual test platform for critical cyberphysical systems in closed-loop. The controller is a fuzzy logic controller, whose structure is designed offline using only the information of the range of the variables, and then, it is online designed in an evolving way, where parameters are adjusted, and new control rules are added based on a novelty detection criterion. The controller is tested on a Two-Tank system in a closed-loop networked environmentunder a proof-of-concept platform for testing cyberphysical systems, named KhronoSim. The proposed self-evolving controller has been successfully evolved/designed, controlling the system on initially unknown regions of operation.