Event‐triggered filtering and intermittent fault detection for time‐varying systems with stochastic parameter uncertainty and sensor saturation

Abstract

Summary In this paper, the event‐triggered filtering and intermittent fault detection problems are investigated for a class of time‐varying systems with stochastic parameter uncertainty and sensor saturation. Due to the existence of event‐triggered mechanism, the measured signal could be transmitted only when it satisfies the triggering condition. An event‐triggered filter is developed, which takes the event‐triggered mechanism, parameter uncertainty, and sensor saturation into full consideration but does not depend on any specific uncertainty structure. By utilizing the inductive and stochastic analysis technique, the filter gain is designed to ensure that the upper bound of the estimation error covariance is minimized at each time step. Based on the proposed filter, a residual is generated and the corresponding evaluation function and detection threshold are given to achieve fault detection. At last, two simulation studies are carried out to demonstrate the effectiveness and applicability of the proposed method.

Distributionally robust trade‐off design of parity relation based fault detection systems

“…• The entropy-based ambiguity set P e (P 0 , 𝜂) includes any distribution P whose KL divergence (i.e., negative relative entropy) with respect to a nominal distribution P 0 is less than or equal to a predefined nonnegative value 𝜂, that is, P e (P 0 , 𝜂) = {P ∈ D|KL(P, P 0 ) ≤ 𝜂} , (11) where D denotes the set of all probability distributions, and KL(P, P 0 ) = ∫ log dP dP 0 dP represents the KL divergence from P to P 0 . In this paper, assume z 0,k ∼ P e (P z 0 , 𝜂) in the fault-free case, that is, the true distribution  z 0 of the fault-free primary residual z 0,k satisfies KL( z 0 , P z 0 ) ≤ 𝜂, with the nominal distribution P z 0 being  (0, Σ 0 ).…”

Fault estimation for discrete time‐variant systems subject to actuator and sensor saturations

“…Here, denotes the i th element of the saturation level . Accordingly, with subscripts u and y , namely, and are, respectively, saturation functions of the actuator and the sensor, and for more details on modeling of the saturated phenomenon on actuator and sensor, please refer to Reference 27,28,32‐37 and references therein. Moreover, without loss of generality, the following assumptions are given for (1):…”

“…If one applies a FD method without considering actuator/sensor saturation, degradation on fault diagnosis performance may emerge. For FE issue, to obtain high accuracy of estimation, these saturations are obliged to be carefully handled simultaneously, and some results are dedicated 32‐37 . It is worth pointing out that, most of FD/FE results on systems with saturation nonlinearities focus on sensor saturation, and few works consider saturated control input.…”

Finite‐horizon fault estimation for time‐varying systems with multiple fading measurements under torus‐event–based protocols

“…In this regard, a natural yet efficient way is used, which is a scheduling strategy to determine whether the data need to be transmitted or not . Consequently, the event‐based transmission has emerged at the right moment . In addition, the issues of event‐based control/filtering have given rise to some research interests and have been studied in other works .…”

“…11,12 Consequently, the event-based transmission has emerged at the right moment. [13][14][15][16][17][18] In addition, the issues of event-based control/filtering have given rise to some research interests and have been studied in other works. [19][20][21][22][23][24][25][26][27] In a typical event-based case, a transmission is triggered only when certain specified events occur, that is, the trigger function is greater than a given threshold that indicates the well-studied unilateral triggering.…”