Hamza BENADADA will defend his PhD on Sept. 9st, 2026 at 9:30AM.
Place : Amphitheatre Clémence Royer, Building 321 Jacqueline Ferrand, Dept. GM, INSA Lyon, Villeurbanne
System identification and observers with unknown inputs for dynamical systems : Application to vehicle’s center of gravity estimation
Jury :
Rapporteurs:
M. Olivier SENAME, Professeur des Universités, Universités Grenoble INP / UGA
M. Guillaume MERCÈRE, Professeur des Universités, Université de Poitiers
Examinatrice:
Mme Marion GILSON, Professeure des Universités, Université de Lorraine
Encadrement :
M. Paolo MASSIONI, Maître de conférences HdR, INSA Lyon, directeur de thèse
M. Michaël DI LORETO, Professeur des Universités, Université Lyon 1, co-encadrant
M. Damien ÉBÉRARD, Maître de conférences, INSA Lyon, co-encadrant
Invité :
M. Fabrice RANC, Ingénieur Sénior, Volvo Group
Abstract :
This thesis focuses on the reconstruction of unmeasured information in dynamical systems from mathematical models and embedded measurements. The work is motivated by an application to the automotive field, and more particularly to the estimation of the load distribution of heavy vehicles.
The first axis deals with the identification of parameters applied to the estimation of the position of the centre of gravity (CG) of a vehicle. This is a key parameter for vehicle safety and dynamics, but it varies with loading and generally cannot be measured directly. The proposed methods rely on vertical vehicle dynamics as well as accelerometer and gyro measurements to identify the position of the CG. They enable this quantity to be estimated without requiring prior knowledge of vehicle parameters other than wheelbase and track width. These methods have a low computational cost, which makes them compatible with an embedded real-time implementation.
The second axis concerns the real-time estimation of unmeasured signals for dynamical systems. This issue is studied in the general framework of observers for linear systems with unknown inputs, whose objective is to reconstruct the inputs and unmeasured states from the only available outputs. The thesis establishes theoretical results on the existence and characterization of these observers, then proposes new observer designs for the state and input reconstruction with constraints of causality and stability. This work provides a unified framework for the inversion of dynamical systems and opens perspectives in areas such as diagnostics, supervision and control of systems subject to unmeasured perturbations.
Keywords: System identification, Vehicle dynamics, Center of gravity position, Observer with unknown inputs, Input observer, Left inversion