Sensor-fault tolerance using robust MPC with set-based state estimation and active fault isolation

Feng Xu (1) , Sorin Olaru (2, 3) , Vicenc Puig (1) , Carlos Ocampo-Martinez (1) , Silviu-Iulian Niculescu (2, 3)

1 UPC - Universitat Politècnica de Catalunya = Université polytechnique de Catalogne [Barcelona]
2 L2S - Laboratoire des signaux et systèmes
3 DISCO - Dynamical Interconnected Systems in COmplex Environments

In this paper, a sensor fault-tolerant control (FTC) scheme using robust model predictive control (MPC) and set-theoretic fault detection and isolation (FDI) is proposed. The MPC controller is used to both robustly control the plant and actively guarantee fault isolation (FI). In this scheme, fault detection (FD) is passive by interval observers, while fault isolation (FI) is active by MPC. The advantage of the proposed approach consists in using MPC to actively decouple the effect of sensor faults on the outputs such that one output component only corresponds to one sensor fault in terms of FI, which can utilize the feature of sensor faults for FI. A numerical example is used to illustrate the effectiveness of the proposed scheme.

Domaines: Automatique / Robotique

Feng Xu, Sorin Olaru, Vicenc Puig, Carlos Ocampo-Martinez, Silviu-Iulian Niculescu. Sensor-fault tolerance using robust MPC with set-based state estimation and active fault isolation. International Journal of Robust and Nonlinear Control, 2017, 27 (8), pp.1260 - 1283. ⟨10.1002/rnc.3627⟩. ⟨hal-01512371⟩