Morris Fuller M.Sc.
Raum: S3|21 201
Development and implementation of Model Predictive Control strategies for test systems handling e-components in powertrains
HiPERFORM: High performant Wide Band Gap Power Electronics for Reliable, energy efficient drivetrains and Optimization through Multi-physics simulation. It aims for a resource-efficient and decarbonized transportation system, supported by the use of advanced and highly integrated wide-bandgap (WBG) technologies in electronic power circuits of electrified vehicles and charging infrastructures.
Contribution of TU Darmstadt
System level modelling and development of optimal model-based control strategies for test systems handling e-components in powertrains: batteries/e-motors and rectifier/inverter.
The current system under investigation consists of two interleaved inverters with fast SiC-switches that control an induction machine.
The interleaving states result in additional space vectors available for machine control but also result in further constraints. Nonlinearities (saturation, torque generation) have to be considered as well.
Model Predictive Control
Based on a finite set of inputs (FCS) and the plant model, predictions of future states of the plant can be made. Minimizing a cost function leads to an optimal control action. The cost function may consist of multiple control aims, e.g. torque and flux accuracy, dc-link ripple etc. An online optimization allows nonlinearities and constraints without approximation.
The realtime capable implementation of the algorithm on SoC-platforms needs tailored modelling (complexity vs. accuracy). Furthermore efficient observer design is necessary. Optimal cost functions have to be found for various control aims.
This project in cooperation with AVL List GmbH has received funding from the ECSEL Joint Undertaking (JU) under grant agreement No 783174. The JU receives support from the European Union’s Horizon 2020 research and innovation programme and Austria, Spain, Belgium, Germany, Slovakia, Italy, Netherlands, Slovenia.