Controle de sistemas a eventos discretos com suporte a otimizações

Abstract

Autonomous systems are increasingly present in the industry, seeking to efficiently perform tasks that are befored manually. In order to maximize production and profit, autonomous systems are expected to be optimized and rigorously calculated, so that they can be flexible enough to support easy parameters changing. Discrete Event Systems (DESs) are used in the industry for process modeling, among them, the autonomous systems. One of the possibilities for DESs control is by using Supervisory Control Theory (SCT). SCT offers a methodology for synthesizing controllable, non-blocking and minimally restrictive controllers for DES. Despite its advantages, optimization aspects are not easily implemented using SCT. Moreover, any change in the system plant imply in synthesizing a new controller, which tends to be expensive both in terms of time or computation. Therefore, although the SCT is safe with respect to control specifications, it does not offer intrinsic optimization techniques, not it behaves flexibly under parameter changes. This work aims to extend the traditional control mesh, keeping its safety benefits, but adding to it a more efficient and flexible feature. For that, we separate the original alphabet into a new set of events, the optimizable events. Based on this event set and on the control action calculated by the classical SCT, a new entity called the Optimizer is able to promote improvements on the control action, thus optimizing the industrial plants under the action of the new controller. After the Optimizer is added, an extended control loop is created. The new control mesh was tested using a practical example, integrating the SCT with a greedy algorithm. It was observed that the proposed method offers improvements in relation to the optimizations when compared with traditional SCT. In addition, the overhead caused by the addition of the Optimizer is not large enough to compromise the control action.