Input design for guaranteed fault diagnosis using zonotopes

An input design method is presented for guaranteeing the diagnosability of faults from the outputs of a system. Faults are modeled by discrete switches between linear models with bounded disturbances and measurement errors. Zonotopes are used to efficiently characterize the set of inputs that are guaranteed to lead to outputs that are consistent with at most one fault scenario. Provided that this set is nonempty, an element is then chosen that is minimally harmful with respect to other control objectives. This approach leads to a nonconvex optimization problem, but is shown to be equivalent to a mixed-integer quadratic program that can be solved efficiently. Methods are given for reducing the complexity of this program, including an observer-based method that drastically reduces the number of binary variables when many sampling times are required for diagnosis.     

Actuator-fault detection and isolation based on set-theoretic approaches

In this paper, an actuator-fault detection and isolation (FDI) approach is proposed. The FDI approach is based on a bank of interval observers, each of which is designed to match a healthy or faulty system mode. To ensure reliable FDI for all considered actuator faults, a collection of invariant set-based FDI conditions are established for the proposed technique. Under these guaranteed FDI conditions, all the considered faults can be detected and isolated during the transition induced by fault occurrences. Comparing with the existing set-based FDI approaches, the advantage of the proposed technique consists in that it combines the advantages of interval observers in the transient-state functioning and the advantages of invariant sets in the steady-state functioning. This paper is completed with the study of a continuous stirred-tank reactor (CSTR), which illustrates the effectiveness of the proposed method.     

FD-ZKF: A Zonotopic Kalman Filter optimizing fault detection rather than state estimation

Enhancing the sensitivity to faults with respect to disturbances, rather than optimizing the precision of the state estimation using Kalman Filters (KF) is the subject of this paper. The stochastic paradigm (KF) is based on minimizing the trace of the state estimation error covariance. In the context of the bounded-error paradigm using Zonotopic Kalman Filters (ZKF), this is analog to minimize the covariation trace. From this analogy and keeping a similar observer-based structure as in ZKF, a criterion jointly inspired by set-membership approaches and approximate decoupling techniques coming from parity-space residual generation is proposed. Its on-line maximization provides an optimal time-varying observer gain leading to the so-called FD-ZKF filter that allows enhancing the fault detection properties. The characterization of minimum detectable fault magnitude is done based on a sensitivity analysis. The effect of the uncertainty is addressed using a set-membership approach and a zonotopic representation reducing set operations to simple matrix calculations. A case study based on a quadruple-tank system is used both to illustrate and compare the effectiveness of the results obtained from the FD-ZKF approach compared to a pure ZKF approach.