A generalized smith predictor for unstable time-delay SISO systems

In this work, a generalization of the Smith Predictor (SP) is proposed to control linear time-invariant (LTI) time-delay single-input single-output (SISO) systems. Similarly to the SP, the combination of any stabilizing output-feedback controller for the delay-free system with the proposed predictor leads to a stabilizing controller for the delayed system. Furthermore, the tracking performance and the steady-state disturbance rejection capabilities of the equivalent delay-free loop are preserved. In order to place this contribution in context, some modifications of the SP are revisited and recast under the same structure. The features of the proposed scheme are illustrated through simulations, showing a comparison with respect to the corresponding delay-free loop, which is here considered to be the ideal scenario. In order to emphasize the feasibility of this approach, a successful experimental implementation in a laboratory platform is also reported. •A generalization of the original Smith Predictor is proposed.•The primary controller can be tuned based on the ideal (delay-free) loop.•The predictor can be adjusted to preserve the steady-state disturbance rejection capabilities of the ideal loop.•A single parameter can be tuned to reach a trade-off between disturbance rejection and robustness.•The strategy can be applied to stable/unstable minimum/non-minimum phase systems