Embedded Nonlinear Optimization Solver for Vehicle Energy Management

Abstract

This thesis investigates real-time embedded solution methodologies for intelligent charge- and trip-planning in electric vehicles, focusing on three core subproblems: optimal driving and charging, eco-driving, and battery thermal tracking. Formulated as constrained nonlinear optimal control problems, these tasks present significant computational challenges for hardware deployment. To balance numerical performance with modeling flexibility, this work explores and contrasts two distinct paradigms: a highly specialized approach based on Pontryagin’s Maximum Principle and a versatile, solver-based Sequential Quadratic Programming framework using acados. Both methods are deployed and validated on a dSPACE MicroAuto- Box II platform. The results demonstrate their real-time feasibility and outline the trade-offs between analytical customization and general optimization frameworks for embedded electric mobility applications.