Robust fault detection for Uncertain Unknown Inputs LPV system

Abstract

International audienceThis paper focuses on robust fault residual generation for Uncertain Unknown Inputs Linear Parameter Varying (U-LPV) systems. Firstly, the problem is addressed in standard LPV systems based on the adaptation of the parity-space approach. The main objective of this approach is to design a scheduled parity matrix according to the scheduling parameters. It results a perfectly decoupled parity matrix face to the system states. Then, the major contribution of this paper relies on the extension to U-LPV systems. Since most of models which represent practical/real systems are subject to parameters variation, unmodeled dynamics and unknown inputs, the approach is clearly justified. The residual synthesis is rewritten in terms of a new optimization problem and solved using Linear Matrix Inequalities (LMIs) techniques. An applicative illustration is proposed and rests on a vehicle lateral dynamic system

Discrete robust switched H∞ tracking state feedback controllers for lateral vehicle control

“…For example, as mentioned previously in Menhour, Koenig, and d'Andréa-Novel (2013a,b), longitudinal velocity and tire cornering stiffnesses are uncertain and vary over the time. Therefore, several solutions based on fuzzy systems (Chadli, Hajjaji, & Oudghiri, 2008;Manceur & Menhour, 2013;Menhour, Manceur, & Bouibed, 2012) and LPV systems (Fergani, Menhour, Sename, Dugard, & d'Andréa-Novel, 2013;Poussot-Vassal et al, 2011;Varrier, Koenig, & Martinez, 2014) are used to overcome such constraints. Moreover, the switched systems appears to be a powerful tool and good solution to deal with such a problem.…”

Fault diagnosis for vehicle lateral dynamics with robust threshold

“…This is because residuals of fault diagnosis systems under normal case are not zero in real application, since there are always some mismatches between the real plant and the mathematical model used for observer design, such as system parameter uncertainties, external disturbances and sensor noises [10], [11]. For example, vehicle lateral dynamics depend on longitudinal velocity [3], [4], [12], which is uncertain due to measurement noises. Besides, it is also subject to bank angle, which can not be easily obtained and thus be treated as mismatched disturbances [13], [14].…”

“…Substituting the values of the rest of parameters [3], one can obtain the following compact form of vehicle lateral dynamiċ …”

“…In addition, to improve response time, safety and save weight, drive-by-wire systems have been developed, where conventional mechanical or hydraulic linkage has been gradually replaced by electronic or electromechanical systems. As a result, modern systems are becoming more complex and safety-critical, since the malfunction of components (e.g., sensors, actuators) may result in dangerous consequences [3].Vehicle lateral dynamics play a key role in vehicle maneuver-ability, stability and driving safety [2], [4], as a result, a lot of Fault Diagnosis (FD) algorithms have been proposed (see, [3]- [5]) to monitor actuator's or sensors' status.…”

The cross-coupling of lateral-longitudinal vehicle dynamics: Towards decentralized Fault-Tolerant Control Schemes

“…In particular, a degraded-mode lateral control for an automated highway system was designed using feedback linearization and a mismatched 3 observer synthesized with H ∞ . Alternatively, in [11,12] a Linear Parameter Varying (LPV ) BM which considers the velocity as a varying parameter was used for the design of a steering controller.…”