Numerical methods for ellipsoid based and tree sparse robust MPC formulations

该项目提案是与亚琛工业大学、柏林工业大学和多特蒙德工业大学的研究小组共同提出的一揽子提案的一部分，涉及鲁棒模型预测控制（MPC）的有效数值方法。主要目的是扩展基于树和椭球体传播的鲁棒MPC方法的最新数值方法，并在高效并行的开源软件中实现和组合这些方法。这项工作可以建立在现有的MPC方法和研究小组开发的软件的基础上，这些方法将通过新的鲁棒公式算法进行扩展，这些公式利用了这些公式所需要的特定结构和并行化的可能性。项目的一个重要组成部分是对实际鲁棒MPC应用程序的包方案进行协调，收集和精确的数学公式，并在仿真中对这些模型进行新的数值方法的测试和评估。

This project proposal which is part of a package proposal together with research groups at RWTH Aachen, TU Berlin and TU Dortmund concerns efficient numerical methods for robust model predictive control (MPC).The principal aim is to extend the state-of-the-art in numerical methods for tree-based and ellipsoidal-propagation-based robust MPC methods, and to implement and combine these methods in efficient and parallel open-source software.The work can build on existing MPC methods and software developed in the research group, which shall be extended by new algorithms for robust formulations, which exploit the specific structures and possibilities for parallelization that these formulations entail.An important part of project is the coordination of the package proposal and the collection and exact mathematical formulation of models of real-world robust MPC applications and to test and evaluate the new numerical methods on those models in simulation.