No abstract

Abstract| This paper deals with the synthesis of greatest linear causal feedback for Discrete Event Systems whose behavior is described in dioid. Such a feedback delays as far as possible the input of the system while keeping the same transfer relation between the input and the output. When a feedback exists in the system, we show how to compute a greater one without decreasing the system's performance.

A behavioral framework for control of (max,+) automata is proposed. It is based on behaviors (formal power series) and a generalized version of the Hadamard product, which is the behavior of a generalized tensor product of the plant and controller (max,+) automata in their linear representations. In the tensor product and the Hadamard product, the uncontrollable events that can neither be disabled nor delayed are distinguished. Supervisory control of (max,+) automata is then studied using residuation theory applied to our generalization of the Hadamard product of formal power series. This yields a notion of controllability of formal power series as well as (max,+)-counterparts of supremal controllable languages. Finally, rationality as an equivalent condition to realizability of the resulting controller series is discussed together with hints on future use of this approach.

This paper deals with just in time control of (max, +)-linear systems. The output tracking problem, considered in previous studies, is generalized by considering additional constraints in the control objective. The problem is formulated as an extremal fixed point computation. This control is applied to timetables computation for urban bus networks.

No abstract