Thesis Work: Online model adaptation for optimal control of HEVs i Göteborg

At Research and Development you will be a key contributor to the next generation outstanding luxury cars from Volvo. Together with other engineers around the world, you and your team will create innovative human-centric car technology that makes life less complicated and more enjoyable for people. Are you interested in design and connected car technology? Do you share our passion for people, the environment and our urge to create a superior driving experience? Research and Development is the place for you to prosper.

Background

Hybrid electric vehicles have one extra degree of freedom compared to conventional vehicles, since they offer the freedom of selecting which actuator (electric or combustion) to use to fulfil the driver’s request. Model based optimal control uses knowledge of the system to optimize the usage of the available actuators, with respect to both efficiency and capability. In doing so it becomes heavily reliant on that the models used accurately capture all important phenomena. If the model quality is low it will have a direct impact on the optimality of the resulting control. The modelled quantities are measured or estimated either directly or indirectly, however only for the current operating point, whereas a model for optimal control needs to capture all reachable operating points.

Scope

The aim of this master thesis is to develop and investigate algorithms for adapting plant and actuator models online. The algorithm should adapt the pre-calibrated models based on measurements/estimates of the modelled quantities, to compensate for model errors, vehicle to vehicle variations, and un-modelled dependencies. This is then to be used to improve the performance of optimal control methods in the vehicles. Since the algorithm should run in real-time both efficiency and robustness needs to be considered.

Profile

• Engineering physics/mathematics/electrical/mechatronics, M.Sc., or similar.
• Knowledge in control theory, optimization, optimal control and mathematical modelling.
• Knowledge in MATLAB. Knowledge in C and Python programming is meritorious.
• Analytical and independent.

Duration

• 20 weeks / 30 ECTS
• Starting date: November-February (Flexible)
• Estimated end date: Summer 2018
• Number of students: 1 or 2 students.
• The work will be performed at dept. 97520 “Vehicle Propulsion, Strategy & Concept”, Volvo Cars Corporation, Göteborg.

Application

• Please include a CV and cover letter with your application as soon as possible but no later than 2017-11-15.
• Selection will be ongoing during the application period.