Analysis und Synthesis of Robustly Controlled Smart-Grid-Systems

能源部门目前正处于从集中式常规发电向分散发电过渡的过程中，其中可再生能源占相当大的份额。由于发电机组数量的增加、运行条件的变化以及可再生能源生产的不确定预测，这种模式的转变增加了未来电力系统设计、管理和控制的复杂性。由于智能电网通常由具有不确定输入和参数的复杂、非线性、微分代数模型描述，因此控制具有挑战性。我们提出了新的控制和分析方法，以增加智能电网的运行区域，从而最大限度地利用可能的智能电网性能，同时在运行条件和不确定性变化日益增加的情况下保持电网的稳定性。为了应对这些挑战，将开发组合技术，通过单独控制和分析子系统来保证全局属性，同时不忽略相互依赖关系或假设它们是静态的。对于电力系统控制，提出了一种结合分层和分布式鲁棒控制原理的方案；对于上层和协调控制层，研究了基于抽象网格模型的预测控制，以实现在线优化负荷分配。在下层，对网格子系统离线合成局部反馈控制器。为了实现这一层的鲁棒性，将研究对具有子系统耦合显式表示的线性参数变系统使用半确定规划。除了两层控制方案的新颖性外，由于输电系统中突然故障引起的切换事件将被考虑并扩展到电网控制的最新技术。提出的控制器综合方法通过开创性的形式化方法进行验证，以保证电网在给定的动态模型和模型不确定性下的稳定性。传统的基于仿真的技术不能证明所有可能的行为都满足给定的规范，因为只能测试无限多可能行为的有限子集。然而，新提出的方法可以通过可达性分析从数学上证明所有规范都是满足的。正式方法还可以提供严格的条件，在这些条件下，经过验证的子系统的集合符合全球规范，从而使与电力系统工业相关的系统尺寸的应用成为可能。所提出的方法将扩展控制和分析非线性微分代数和常微分方程的大型分布式系统的最新技术，也超出了电力系统的领域。

The energy sector is currently in the midst of a transition from centralized conventional power generation towards decentralized generation with a considerable share of renewable energy sources. This paradigm shift increases the complexity of the design, management, and control of the future power system due to the increase in the number of generating units, changing operating conditions, and uncertain prediction of renewable energy production. The control is challenging since smart grids are generally described by complex, nonlinear, differential-algebraic models with uncertain inputs and parameters. We propose new control and analysis methods to increase the operating region of smart grids and thus maximally exploit possible smart grid performance while maintaining grid stability under increasing variations in operating conditions and uncertainties. In order to address these challenges, compositional techniques will be developed to guarantee global properties by controlling and analyzing subsystems separately, while not ignoring the interdependencies or assuming that these are static. For power system control, a scheme combining principles of hierarchical and distributed robust control is envisaged: For an upper and coordinating control layer, the use of predictive control based on abstracted grid models is investigated to achieve an optimized load assignment online. On the lower layer, local feedback controllers are synthesized offline for the grid subsystems. To achieve robustness on this layer, the use of semi-definite programming for linear parameter-varying systems with explicit representation of the subsystem coupling will be investigated. In addition to the novelty arising from the two-layer control scheme, switching events as arising from abrupt faults in the transimission system will be considered and extend the state of the art in grid control.The proposed controller synthesis approach is verified by pioneering formal methods for guaranteeing grid stability in compliance with a given dynamic model and model uncertainties. Traditional simulation-based techniques cannot prove that all possible behaviors meet given specifications, since only a finite subset of infinitely many possible behaviors can be tested. The newly proposed method, however, can mathematically prove that all specifications are satisfied using reachability analysis. The formal approach also makes it possible to provide rigorous conditions under which a collection of verified subsystems meets global specifications, such that the application to system sizes relevant in the power systems industry becomes possible.The proposed methods will extend the state-of-the-art in controlling and analyzing large and distributed systems of nonlinear differential-algebraic and ordinary differential equations also beyond the domain of power systems.