Paulo Andre Nobre Rosa

Fault Detection and Isolation of LTV systems using Set-Valued Observers

This paper introduces the novel concept of using Set-Valued Observers (SVOs) in Fault Detection and Isolation (FDI), for discrete-time linear time-varying systems. The proposed method relies on SVO-based model invalidation to discard models that are not compatible with the input/output data. We argue that there are mainly three significant advantages of using SVOs for FDI, when compared to the most common strategies available in the literature: i) under suitable conditions, we can guarantee that there will be no false alarms; ii) unlike residual-based architectures, the proposed technique does not require the computation of a threshold to declare faults; iii) the SVOs can be used with a wide class of time-varying linear uncertain discrete-time systems. We further show, in simulation, that the proposed FDI algorithm in general requires a very small number of iterations to detect and identify a faulty behavior.

On the distinguishability of discrete linear time-invariant dynamic systems

This paper introduces the notion of absolutely distinguishable discrete dynamic systems, with particular applicability to linear time-invariant (LTI) systems. The motivation for this novel type of distinguishability stems, in particular, from the stability and performance requirements of worst-case adaptive control methodologies. The main results presented herein show that, in most practical cases, a persistence of excitation type of condition and a minimum number of iterations are required to properly distinguish two dynamic systems. We also demonstrate that the former constraint can be written as a lower bound on the intensity of the exogenous disturbances. The applicability of the developed theory is illustrated with a set of examples.

Autolanding Controller for a Fixed Wing Unmanned Air Vehicle

This paper addresses the autolanding guidance and control problem for unmanned autonomous vehicles (UAV) based on the information provided by the onboard Navigation System. The proposed solution relies on a path-following and velocity profile tracking controller synthesized using an accurate aircraft dynamic model, referred to as SymAirDyn, which was designed with the objective of exploiting the whole aircraft’s flight envelope. A suitable polytopic Linear Parameter Varying (LPV) representation with piecewise affine dependence on the parameters is adopted to accurately model the desired aircraft dynamics over a set of predefined operating regions. The synthesis problem is stated as a continuous-time H2 control problem for LPV systems and solved using Linear Matrix Inequalities (LMIs). During the controller design stage, the aircraft landing maneuver is decomposed in the two standard phases: the glideslope and the flare. For each phase, the required control objectives are identified and specific controllers are designed. The controller implementation is tackled within the framework of gain-scheduling control theory using the D-methodology. The performance and effectiveness of the resulting nonlinear gain scheduling controller is illustrated in simulation, using the nonlinear model SymAirDyn under different types of disturbances, namely Dryden spectrum generated wind during the glideslope and wind gusts near touchdown.

A Set-Valued Approach to FDI and FTC of Wind Turbines

A complete methodology to design robust fault detection and isolation (FDI) filters and fault-tolerant control (FTC) schemes for linear parameter varying systems is proposed, with particular focus on its applicability to wind turbines. This paper takes advantage of the recent advances in model falsification using set-valued observers (SVOs) that led to the development of FDI methods for uncertain linear time-varying systems, with promising results in terms of the time required to diagnose faults. An integration of such SVO-based FDI methods with robust control synthesis is described, to deploy new FTC algorithms that are able to stabilize the plant under faulty environments. The FDI and FTC algorithms are assessed by resorting to a publicly available wind turbine benchmark model, using Monte Carlo simulation runs.

Path-Following Control for Coordinated Turn Aircraft Maneuvers

This paper addresses the path-following problem of steering an autonomous airplane along a predefined 3-D path, while performing a coordinated turn maneuver. The presented solution relies on the definition of a path-dependent error space to express the dynamic model of the vehicle, and of an output tracking error that guarantees both path-following and coordinated turn compliance. For controller design purposes, the error dynamics are approximated by a polytopic Linear Parameter Varying (LPV) system representation with piecewise affine dependency on the parameters. The synthesisproblem is stated as an H2 minimization problem with pole placement contraints, and solved using Linear Matrix Inequalities (LMIs). The nonlinear controller is implemented within the scope of gainscheduling control theory, using the D-methodology. The performance of the designed controller is assessed in simulation, using the full nonlinear model of a small scale airplane.

Multiple-model adaptive control with set-valued observers

This paper proposes a multiple-model adaptive control methodology, using set-valued observers (MMAC-SVO) for the identification subsystem, that is able to provide robust stability and performance guarantees for the closed-loop, when the plant, which can be open-loop stable or unstable, has significant parametric uncertainty. We illustrate, with an example, how set-valued observers (SVOs) can be used to select regions of uncertainty for the parameters of the plant. We also discuss some of the most problematic computational shortcomings and numerical issues that arise from the use of this kind of robust estimation methods. The behavior of the proposed control algorithm is demonstrated in simulation.

Fault Detection and Isolation and Fault Tolerant Control of Wind Turbines using Set-Valued Observers

Abstract Research on wind turbine Operations & Maintenance (O&M) procedures is critical to the expansion of Wind Energy Conversion systems (WEC). In order to reduce O&M costs and increase the lifespan of the turbine, we study the application of Set-Valued Observers (SVO) to the problem of Fault Detection and Isolation (FDI) and Fault Tolerant Control (FTC) of wind turbines, by taking advantage of the recent advances in SVO theory for model invalidation. A simple wind turbine model is presented along with possible faulty scenarios. The FDI algorithm is built on top of the described model, taking into account process disturbances, uncertainty and sensor noise. The FTC strategy takes advantage of the proposed FDI algorithm, enabling the controller reconfiguration shortly after fault events. Additionally, a robust controller is designed so as to increase the wind turbines performance during low severity faults. Finally, the FDI algorithm is assessed within a publicly available benchmark model, using Monte-Carlo simulation runs.

A Set-Valued Approach to FDI and FTC: Theory and Implementation Issues

Abstract A complete methodology to design robust Fault Detection and Isolation (FDI) filters and Fault Tolerant Control (FTC) schemes for Linear Time-Varying (LTV) systems is proposed. The paper takes advantage of the recent advances in model invalidation using Set-Valued Observers (SVOs) that led to the development of FDI methods for uncertain linear time-varying systems, with promising results in terms of the time required to diagnose faults. An integration of such SVO-based FDI methods with robust control synthesis is described, in order to deploy new FTC algorithms that are able to stabilize the plant under faulty environments. The FDI algorithm is assessed within a wind turbine benchmark model, using Monte-Carlo simulation runs.

Stability overlay for adaptive control laws applied to linear time-invariant systems

Two broad classes of adaptive control algorithms can be found in the literature: i) stability based, with minimal assumptions on the plant; ii) performance based, with relatively more stringent assumptions on the plant. This paper proposes a solution, referred to as Stability Overlay (SO), to enable stability guarantees in performance based algorithms. In our methodology, the performance based adaptive control laws are only responsible for designating the controller that should be selected; the SO decides whether this controller should or not be used, based upon its most recent history of utilization. We argue that using two algorithms in parallel — the SO for stability purposes and any other suitable for the performance requirements — leads to higher levels of performance while guaranteeing stability of the adaptive closed-loop for bounded (but unknown) disturbances. The SO methodology is applicable to both time-invariant and time-varying, nonlinear and linear systems. However, due to space limitations, we only consider linear time-invariant (LTI) plants in this paper. The theory is illustrated with an example.

Model Falsification of LPV Systems Using Set-Valued Observers

Abstract This paper presents a novel model falsification strategy, based on Set-Valued Observers (SVOs) for uncertain Linear Parameter-Varying (LPV) systems. We show that, under mild conditions, the proposed methodology is able to select the correct model of the plant, among a set of plausible ones. The applicability of this approach is illustrated in simulation, resorting to a single-link robotic arm with a revolute elastic joint, with uncertain spring stiffness.