Algebraic Synthesis for Online Adaptation of Dependable Discrete Control Systems

Abstract Common practice in industrial design of discrete controllers as well as in most synthesis procedures advocated for discrete control in academia is to create the control logic and to transfer it into a PLC language before process start-up. Changes in the operational constraints of the controlled process (e.g. of available resources, nominal set-points, occurrences of failures) have to be accounted for by dedicated alternative routines, i.e. dependability is restricted to variations which are envisaged during design. In contrast, the approach proposed in this paper operates online on a model of the uncontrolled process to compute a control strategy that is adapted to the current set of constraints. By using algebraic computations largely resembling techniques for discrete-time continuous-valued controllers, perceived process variations (including newly defined control-goals) are first assessed with respect to the existence of a feasible successful control strategy, before such a dependable and optimal strategy is computed.