Karim Chabir

Design of an adaptive Kalman filter for fault detection of Networked Control Systems

Networked control systems (NCS) are an active area of research within control and estimation. Networked control systems have many advantages by comparison with the traditional ones. The fault detection problem is studied in this paper for a class of linear networked control systems (NCS) with communication delays. Aim is to generate residual signals which, in the fault free case, are supposed to be identically zero. In practice, this condition is not satisfied due to various factors such as measurements noises, model uncertainties, disturbances and also for NCS, induced delays and data dropouts. In this paper, a new approach for robust fault diagnosis is presented. The method is based on the use of an adaptive Kalman filter designed in order to minimize the effect of the network.

Design of Fault Isolation Filter under network induced delay

The fault detection problem is studied in this paper for a class of linear Networked Control Systems (NCS) with communication delays. Aim is to generate residual signals which, in the fault free case, are supposed to be identically zero. In practice, this condition is not satisfied due to various factors such as measurements noises, model uncertainties, and in particular for NCS communication induced delays. The aim of this paper is to design a Fault Isolation Filter satisfying to directional properties and being robust against network induced delays as well.

Extended Kalman Filter taking account of extra delays and intermittent data in Network Controlled Systems

Abstract Network Controlled Systems are becoming very popular today. However, the use of a network in the control loop induces extra difficulties. Among these difficulties, packet losses and extra delays must be carefully considered. This paper deals with both of these problems. The first one is tackled through an Extended Kalman Filter that is insensitive to data losses. The technique proposed is derived from the Kalman Filter previously proposed by Sinopoli et al. The second difficulty is considered through another Extended Kalman Filter, the state of which is augmented with the delay. The techniques proposed are exemplified on the UAV quadrotor benchmark developed at the GIPSA-Lab.

On Fault Detection and Isolation (FDI) Design for Networked Control Systems with Bounded Delay Constraints

This paper studies the problem by fault detection and isolation in networked control systems with bounded transmission delays caused from the communication network. In the residual generation, the effect of the networked induced bounded delays as well as the noise is reduced. The minimization of false alarm rate caused from unknown inputs as well as the variation of control inputs is achieved in the residual evaluation stage. A residual evaluation technique, which combines the norm-based and statistic schemes, is applied. This includes the design of an adaptive threshold. An illustrating example is given to show the efficiency of the obtained results.

Adaptive thresholding for fault detection in networked control systems

This paper studies the problem of fault detection in networked control systems. In the residual generation, the effect of the networked induced bounded delays as well as the noise is reduced. The minimization of false alarm rate caused from unknown inputs as well as the variation of control inputs is achieved in the residual evaluation stage. This includes the design of an adaptive threshold. An illustrating example is given to show the efficiency of the obtained results.

Robust fault diagnosis of networked control systems via Kalman filtering

The fault detection problem of a class of linear networked control systems (NCS) with communication delays will be studied in this paper. The aim of the study is to generate residual signals which, in the fault free case, are supposed to be identically zero. In practice, this condition is not satisfactory due to various factors such as measurements noise, model uncertainties, and more specifically the NCS communication induced delays. The effect of unknown networked induced delays on conventional observed residue generator is also studied in this paper. It is shown that the detection performances may be deteriorated because of the sensitivity of the residuals to the delays. The proposed approach is based on robust residual generation based on Kalman filtering.

Two-stage Kalman filter for simultaneous fault and delay estimation in networked control systems

Networked Control Systems are becoming very popular nowadays. However, the use of networks in the control loop induces extra difficulties in diagnosis. In this paper we propose a method for estimate fault as well delay induced by the network. Inspired from the Kalman filter previously proposed by Sinopoli et al. and Hsieh et al., two solutions are presented; the augmented and the two-stage Kalman filter. Main attention is paid for the last one. The delay and the fault are integrated as a state variable. It is supposed that the delay and the fault are random variables. In addition, we provide some analytic results to demonstrate the computational advantages of two-stage Kalman filter over augmented ones. An example illustrates the obtained results is given.

State Filtering for Networked Control Systems Subject to Switching Disturbances

Abstract State estimation of stochastic discrete-time linear systems subject to unknown inputs has been widely studied, but few works take into account disturbances switching between unknown inputs and constant biases. We show that such disturbances affect a networked control system subject to deception attacks on the control signals transmitted by the controller to the plant via unreliable networks. This paper proposes to estimate the switching disturbance from an augmented state version of the intermittent unknown input Kalman filter. The sufficient stochastic stability conditions of the obtained filter are established when the arrival binary sequence of data losses follows a Bernoulli random process.