Morteza Rezaei Larijani - Mohammad soutient sa thèse de doctorat lle Lundi 10 juin 2024 à 14h00, à l'École doctorale sciences, technologies, santé (UPJV EDSTS 585) à Amiens (salle des Thèses ED STS - Pôle Scientifique - 33 rue Saint Leu - 80039 AMIENS)
Jury
Prof. Ahmed EL HAJJAJI: Université de Picardie Jules Verne (Directeur)
Dr. Shahin HEDAYATI KIA: Université de Picardie Jules Verne (Co-Directeur)
Dr. Mohammad Reza ZOLGHADRI : Sharif University of Technology (Co-Directeur)
Dr. Amir TAGHAVIPOUR : K. N. Toosi University of Technology (Conseiller)
Prof. Bogdan Marinescu: Ecole Centrale Nantes (Rapporteur)
Prof. Lahoucine ID-KHAJINE : IUT de Cergy-Pontoise (Rapporteur)
Prof. Ebrahim FARJAH : Shiraz University (Examinateur)
Dr. Maryam Babazadeh : Sharif University of Technology (Examinateur)
Prof. Hubert RAZIK: Université de Lyon 1 (Examinateur)
Abstract
Energy management strategies (EMSs) play a crucial role in enhancing the performance of electric vehicles (EVs). This Ph.D. thesis focuses on the development of an intelligent EMS (IESM) aimed at mitigating battery degradation by optimally splitting the load current within a hybrid energy storage system (HESS), which integrates batteries and supercapacitors (SCs) in a connected EV. The battery and SC packs are arranged in a semi-active topology utilizing a bi-directional DC-DC converter, that regulates the SC current through a model predictive control (MPC) supervisory controller. Initially, a straightforward approach is proposed for real-time modeling of the EV including its primary electrical and mechanical parts. To represent the Li-ion battery cell, a double RC equivalent circuit is employed, with model parameters dependent on the state-of-charge (SoC), and is extracted using the electrochemical battery cell model in MapleSim commercial software. Subsequently, additional components such as SCs, the DC-DC converter, inverter, electric motor, and mechanical transmission are incorporated into the model. The resulting EV model is compared with the EV model from CarSim commercial software. For optimization-based control, the state-space representation of the HESS is derived. Given that these equations are linear and parameter-dependent, we propose employing a linear parameter-varying model predictive control (LPV-MPC) approach. This leads to a quadratic programming (QP) optimization problem with a cost function based on the square of the battery pack current and the squared error of the state-of-voltage (SoV) of the SC. Additionally, SoV control is influenced by the EV’s upcoming acceleration, which can be estimated through Vehicule-to-Vehicule (V2V) communication data by adjusting a related weighting factor, thereby providing a better opportunity to extend the battery lifecycle. The proposed LPV-MPC-based IEMS and HESS model are evaluated and implemented in a real-time digital simulator (RTDS) based on dSPACE SCLEXIO under two standard drive cycles. The results are compared with five EMSs: LPV-MPC with fixed SoV control, LPV-MPC with speed-dependent SoV control, LTI-MPC, filter-based method, and rule-based approach. Compared with LPV-MPC with fixed SoV control, the proposed method demonstrates reductions of up to 18.82% in battery current root-mean-square, 30.26% and 25.85% in discharge/charge peak current, 9.71% in ampere-hour throughput, 4.78% and 29.06% in capacity and energy loss, respectively.
