Control of Dynamic Discrete-Event Systems
Lenko Grigorov · 2004
In this work a type of discrete-event systems which vary with time, named dynamic discrete-event systems, is defined and a method for the control of such systems is proposed. Two major topics are discussed: redundancy structures for the efficient reconstruction of the complete system models when constituent modules change, and the use of online control to optimize the supervision of dynamic systems. Three redundancy structures are proposed: stack redundancy, tree redundancy and hybrid redundancy. The control algorithms are designed to address the specific requirements of the class of systems considered. The online control paradigm is used since it can adapt automatically to the changes that occur in dynamic systems. A value function is used to guide the supervision of systems, allowing for optimal control and for a flexible way to set requirements on the expected system behavior. This function, combined with the definition of goals, is shown to successfully replace marking in automata and to support the work of continuous-life systems using infinite goals. It is shown experimentally that the proposed modifications offer a significant improvement in the quality of control of dynamic systems. The results of this work can be applied to the control of large dynamic systems with prioritizing: such as dispatching centers, factories, resource access managers, and others.