Robust and optimal trajectory and powertrain control of intelligent electric vehicles for safe and energy-efficient driving in interactive traffic conditions
File(s)
Author(s)
Yu, Sheng
Type
Thesis or dissertation
Abstract
With the development of electric Connected and Autonomous Vehicle (CAV) technologies, this thesis presents optimal control methods for three representative scenarios: Ecological Adaptive Cruise Control (EACC), autonomous overtaking, and dual-motor energy management, providing robust and optimal trajectory and powertrain control solutions for electrical CAVs.
The first part addresses the EACC problem with two robust Model Predictive Control (MPC) strategies. The time-domain MPC formulation employs feedback linearisation to simplify vehicle dynamics and incorporates a Linear Matrix Inequality (LMI)-based robustness mechanism to handle system disturbances. To further improve ecological performance and computational efficiency, a space-domain Robust EACC (REACC) is proposed, yielding convex vehicle dynamics, accurately fitted electric powertrain models, and real-time implementation capability. Simulation results demonstrate that REACC achieves a 4.5%—5% improvement in energy efficiency, reduces energy loss by approximately 11%, and decreases computation time by at least 17% compared to benchmarks.
The second part investigates autonomous overtaking in mixed traffic, where the surrounding vehicle is an unconnected Human-Driven Vehicle (HDV). A Game-Theoretic PRedictive Overtaking (GT-PRO) strategy is developed to jointly coordinate longitudinal and lateral vehicle motions through two coupled MPCs. A Stackelberg game-based MPC predicts the HDV’s reactions, while a stochastic MPC governs longitudinal control under chance constraints derived from real-world trajectory data. Simulation results demonstrate GT-PRO’s safety, efficiency, comfort, and real-time feasibility, achieving up to approximately a 75% reduction in lateral acceleration compared with benchmarks when the CAV merges back during overtaking.
In addition to trajectory optimisation, this thesis investigates an energy management strategy for dual-motor electric vehicles by considering the temperature effect on motor energy conversion efficiency. After collecting experimental data, a novel motor temperature dynamics model and temperature-dependent motor energy conversion efficiency maps are obtained. Using these realistic results, an optimal energy management solution is determined, achieving battery energy savings of up to 11.96% compared to benchmarks.
The first part addresses the EACC problem with two robust Model Predictive Control (MPC) strategies. The time-domain MPC formulation employs feedback linearisation to simplify vehicle dynamics and incorporates a Linear Matrix Inequality (LMI)-based robustness mechanism to handle system disturbances. To further improve ecological performance and computational efficiency, a space-domain Robust EACC (REACC) is proposed, yielding convex vehicle dynamics, accurately fitted electric powertrain models, and real-time implementation capability. Simulation results demonstrate that REACC achieves a 4.5%—5% improvement in energy efficiency, reduces energy loss by approximately 11%, and decreases computation time by at least 17% compared to benchmarks.
The second part investigates autonomous overtaking in mixed traffic, where the surrounding vehicle is an unconnected Human-Driven Vehicle (HDV). A Game-Theoretic PRedictive Overtaking (GT-PRO) strategy is developed to jointly coordinate longitudinal and lateral vehicle motions through two coupled MPCs. A Stackelberg game-based MPC predicts the HDV’s reactions, while a stochastic MPC governs longitudinal control under chance constraints derived from real-world trajectory data. Simulation results demonstrate GT-PRO’s safety, efficiency, comfort, and real-time feasibility, achieving up to approximately a 75% reduction in lateral acceleration compared with benchmarks when the CAV merges back during overtaking.
In addition to trajectory optimisation, this thesis investigates an energy management strategy for dual-motor electric vehicles by considering the temperature effect on motor energy conversion efficiency. After collecting experimental data, a novel motor temperature dynamics model and temperature-dependent motor energy conversion efficiency maps are obtained. Using these realistic results, an optimal energy management solution is determined, achieving battery energy savings of up to 11.96% compared to benchmarks.
Version
Open Access
Date Issued
2025-08-05
Date Awarded
2026-01-01
Copyright Statement
Attribution-NonCommercial 4.0 International Licence (CC BY-NC)
License URL
Advisor
Simos, Evanglou
Imad, Jaimoukha
Publisher Department
Department of Electrical and Electronic Engineering
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)