In this paper, a sensor stuck fault-tolerant control framework for linear time-invariant plant models subject to input/state constraints and bounded disturbances is presented. A receding horizon control reconfigurable scheme is proposed to contrast undesired effects due to sensors malfunctioning. The main merit of this strategy relies on its intrinsic capability to quickly identify fault occurrences and to take a decision on the adequate control action. This is formally obtained by jointly exploiting set-theoretic polyhedral ideas and the certainty equivalence concept. A numerical example is provided and the control performance contrasted with a well-reputed competitor fault-tolerant control scheme.
KEYWORDSconstrained systems, fault-tolerant control, piecewise affine models, set invariance, stuck sensors
INTRODUCTIONFault-tolerant control (FTC) is an idea well known in the research literature, 1,2 which allows to hold plant performance close to a desirable state and to preserve stability conditions in the presence of component and/or instrument malfunctions. Accommodation capability is based on several aspects like fault severity, plant robustness, and mechanisms inducing redundancy in sensors and/or actuators. As a consequence, FTC techniques are an appealing methodology due to continuously growing demand of safety and reliability in several technical applications. 3 The literature considers two main groups of FTC approaches, ie, passive and active schemes (see the work of Patton 4 for a review). Passive FTC consists of control laws considering fault occurrence as a system perturbation. As a consequence and within certain margins, the control law, kept fixed, has inherent fault-tolerant capabilities, allowing the system to cope with the fault presence. On the other hand, active FTC techniques consist in adapting the control law using the information given by a model-based fault detection and isolation (FDI) software module. 5 By means of this information, self-tuning in the regulation loop are accomplished after fault appearance trying to satisfy the control objectives with minimum performance degradation.Here, an active reconfiguration control strategy for plants described via discrete-time piecewise affine systems is presented. In particular, we focus our attention on sensor devices, necessary to acquire real-time system operating information. Sensors are thought to break down as frequently as actuators, and sensor failures often bring serious and even disastrous situations. Faulty sensors may in fact produce performance degradation, process shutdown, or a fatal accident. When process operation elements, controllers, and actuators are functional, the undesirable consequences of a faulty sensor can be prevented by using appropriate FDI methodologies. In particular, an FDI sensor-based technique allows to detect faulty sensors and the identification of the fault type (bias, drift, noise, or complete failure). This FDI category is a special case in the general FDI field that recently has received extensive atten...