Controle supervisório cooperativo empregando a meta-heurística sistema de formigas

Abstract

The modern industry contemplates a profile of systems that detect and react to – in time for execution – changes in the physical context, operating in different ways. Even though these characteristics show potential to add improvements to the productive processes, they also pose challenges to the Control and Automation Engineering community, such as the limited capacity to process, in parallel, multiple and complex control loops. An alternative appears from the combination of formal methods of synthesis, such as the Supervisory Control Theory (SCT), with architectures that decentralize the implementation of the control system. That benefits both the programming and the implementation of the controller. However, current implementation models don’t consider emerging aspects of the industry, such as intelligent controllers, for example, which is something that shows potential for efficient resources usage, production customization and cooperative production. Therefore, to achieve this scope, this work proposes the making of a smart controller from the integration of optimization meta-heuristics to the control system derived from SCT. The state’s space from the result of the control system represents all possible sequences that are operationally safe. Thus, the central idea found in this research comes from the use of the mete-heuristic Ant System (AS), which explores this space building chains of events that, when enabled, control the behavior of physical devices on the factory floor. In this approach, the meta-heuristics play a key role in building solutions that create working cooperation between devices and minimizes production costs due to stochastic demands that may occur. Another key aspect of this work comes from the use of a communicating control architecture, which brings the concept of a global controller as well as several local-agents controllers. The global controller performs an abstraction of the behavior of the local agents, being responsible for the high-level coordination of the control system logic, which results in the local physical actions. This abstraction allows the acquirement of a simpler, more compact global controller, reducing the searching as well as the complexity of the combinatorial issue and computational load, bringing benefits related to the performance of the implemented meta-heuristics. In terms of results, the observed data brought improvements to the models used and added great potential regarding cost reduction and work cooperation between devices, through the application of meta-heuristic methods integrated to a monolithic supervisory control system, decentralized or not. in addition, other aspects of implementation indicate high potential for scalability, which implies possible gains for TCS, with the possibility of expansion to various researches and use in productive environments.