Atrasos variantes no tempo realmente degradam o desempenho de um sistema de controle em rede?

Abstract

Time varying delays really degrade the performance of a networked control system? Point of view of sampled data theory for linear time-invariant system is no doubt that time-invariant delays degrade the closed loop performance of linear feedback control systems. On the other hand, it is not obvious that the performance of a linear or nonlinear feedback control system will be degraded when the sampling rate and the time delay are both time-varying. Time-varying sampling rates and time-varying delays appear quite often in a networked control system where time-driven (sensing signals) and event-driven (control actions) paradigms coexist. This thesis discusses the design of feedback control law where these two paradigms are combined and thus, there is no performance degradation of the closed loop system. The document shows that by using the proposed control law we improve the performance like visible reduction of overshoot, rise time and settling time. These facts can be interpreted as improvements of the feedback control system performance. The theoretical reasons of these improvements are elucidated and proved and are contributions of this thesis. At the end, a new level-crossing sampling scheme was developed to make feasible and to prove the stability of the proposed networked control systems in an easy way with accuracy and in a non-conservative way. This new scheme is able to sample a signal with time-varying sampling intervals. Besides, this new sampling scheme can be used with SISO and MIMO systems, time-varying or not, time- or event-driven, or yet under combinations of both time and events. Lastly, several simulated and experimental examples are shown and the results of those are used to elucidate the mentioned advantages as contributions of this thesis.