Fault tolerant control allocation using unknown input observers

Fault-tolerant control allocation with actuator dynamics: Finite-time control reconfiguration

“…Such problems can be avoided by accommodating the fault effects if a suitable control reconfiguration policy is considered. Moreover, it can be noticed that the case of faults affecting the actuator input v(t), as well as actuator total failures, can be tackled using the framework proposed in [9]. Summarizing, throughout the paper we will deal with the system resulting from the coupling of (1), (2), (5) and (8) with desired control effect assigned by (4).…”

“…Mimicking the approaches presented in [8] and [9], a family of unknown input observers for system (1) can be designed in order to detect and isolate the faults affecting the actuator dynamics (8). To this purpose, with respect to the matrix W = BG ∈ R n×m , we call uniform sub-rank the integer k 0 ≤ k computed as follows:…”

“…The threshold (h) can be computed or tuned based on known bounds for observer initialization error as well as for possible disturbance or noise terms affecting the plant [9].…”

“…However, one key point for successfully re-allocating the control is the availability of adequate information about the faults that have occurred; indeed, some accurate fault estimation and/or a correct identification of the faulty actuators or effectors are necessary to address the reconfiguration. Recent results toward fault tolerant control allocation are based on slidingmode techniques [7] [14], adaptive control strategies [5] [18] [19] and unknown input observers [8] [9]. Further investigations on this topic, with a more application-oriented character, are proposed for reconfiguration in flight control [4], [22], [20] and fault accommodation in automated underwater vehicles [17].…”

“…In Section II the basic setup of control allocation with actuator dynamics is introduced, while Section III is devoted to the presentation of the proposed method for fault detection/isolation based on unknown input observers, assuming that the actuator state is not directly measured; for sake of simplicity, the output of the plant is assumed to be the whole state. However, such assumption can be relaxed and methods proposed in [9] can be used to tackle the case of systems with unmeasured states and inputs. Section IV reports the main results of the paper: first the steady-state effects of the faults are estimated, then a finite-time actuator reconfiguration strategy is defined and finally the fault accommodation of the overall system is proved.…”