Reusable liquid-propellant rocket engines (LPREs) imply more demanding robustness requirements than expendable ones due to their extended capabilities. Therefore, the goal of this work was to develop a control loop adapted to all the operating phases of LPRE, including transients, and robust to internal parametric variations. Firstly, thermo-fluid-dynamic simulators representative of the gas-generator-cycle engines were built. These simulators were subsequently translated into nonlinear state-space models. Based on these models, the continuous subphase of the start-up transient is controlled to track precomputed reference trajectories. Beyond the start-up, throttling scenarios are managed with end-state-tracking algorithm. Model Predictive Control has been applied in a linearised manner with robustness considerations to both scenarios, in which a set of hard state and control constraints must be respected. Tracking of pressure (thrust) and mixture-ratio operating points within the design envelope is achieved in simulation while respecting constraints. Robustness to variations in the predominant parameters, to external state perturbations and to the possible impact of an observer on the loop, is demonstrated.

中文翻译：

---

基于模型的可重复使用液体推进剂火箭发动机的鲁棒瞬态控制

可重复使用的液体推进剂火箭发动机 (LPRE) 由于其扩展能力而意味着比消耗性发动机更苛刻的稳健性要求。因此，这项工作的目标是开发一个适用于 LPRE 的所有操作阶段（包括瞬态）并且对内部参数变化具有鲁棒性的控制回路。首先，建立了代表燃气发电机循环发动机的热流体动力学模拟器。这些模拟器随后被转化为非线性状态空间模型。基于这些模型，启动瞬态的连续子阶段被控制以跟踪预先计算的参考轨迹。除了启动之外，节流场景还通过最终状态跟踪算法进行管理。模型预测控制已以线性化的方式应用，并考虑到两种情况的稳健性，其中必须遵守一组硬状态和控制约束。设计范围内的压力（推力）和混合比操作点的跟踪是在模拟中实现的，同时尊重约束。证明了对主要参数变化、外部状态扰动和观察者对回路的可能影响的鲁棒性。