Multiscale-control of the low-temperature combustion process GCAI

最先进的方法闭环控制的低温燃烧过程GCAI是基于循环到循环的反馈控制。然而，这些方法只允许在非常有限的引擎映射中稳定运行。在基于周期到周期的控制中，只有在同一时间尺度上发生的动态影响和干扰才能被控制。决定低温燃烧的稳定性和排放特性并在内循环时间水平上运行的相关物理化学过程是无法控制的。出于这个原因，TP1研究了考虑较小时间尺度的多尺度控制算法。通过对这些关键时间尺度的成功控制，预计在提高效率和减少污染物排放的同时，运营地图将显着扩大。作为第一个资助期的一部分，开发并实施了多尺度控制方法的具体概念。在TP1中，针对负载和速度瞬态运行场景，讨论了多尺度控制的面向控制扩展。为了明确考虑发动机转速，我们打算扩展现有的模型，以包括转速变化的影响。为此，与TP3合作在发动机试验台进行了新的试验，并根据分析的效果对模型进行了扩展。此外，将研究小组开发的反应动力学整合到缩短时间模型中，以提高预测和相关控制质量。在这方面，将特别审查模型降阶技术。为了解决重大的过程不确定性，将研究以物理驱动干扰模型的形式考虑进一步的干扰。此外，还将研究鲁棒控制方法。虽然TP1侧重于相应的系统特定问题公式，特别是对不确定性的充分描述，但开发的算法将在TP2中进行扩展，以满足高计算需求，尽管鲁棒方法会增加操作。最后，将所开发的控制算法与TP3在发动机试验台上进行了验证。决定性的标准是覆盖的运行图，其中可以实现稳定运行，同时考虑到瞬态负载和速度分布，以及减少排放和提高效率的潜力。

State-of-the-art approaches for closed-loop control of the low temperature combustion process GCAI are based on cycle-to-cycle feedback control. However, these approaches allow only a stable operation in a very limited engine map. With cycle-to-cycle based control, only the dynamic effects and disturbances that occur on the same time scale can be controlled. The relevant physicochemical processes that determine the stability and emission characteristics of low temperature combustion and run on an inner-cyclic time-level, cannot be controlled. For this reason TP1 investigates multiscale control algorithms that account the smaller time scales. It is expected that with successful control on these critical time scales, the operating map will be significantly expanded while improving efficiency and reducing pollutant emissions. As part of the first funding period, a concrete concept for the multi-scale control approach was developed and implemented.In TP1, the control-oriented expansion of the multi-scale control with regard to load- and speedtransient operating scenarios is addressed. In order to take the engine speed explicitly into account, it is intended to expand the current models to include the effects of changing speed. For this purpose, new experiments are carried out on the engine test bench in cooperation with TP3 and the models are extended based on the analysed effects. In addition, the integration of the reaction kinetics, which was developed in the research unit, into the reduced-time models to improve prediction and the associated control quality will be investigated. In this regard, particularly model order reduction techniques will be examined.To tackle the significant process uncertainty, consideration of further disturbances in form ofphysically motivated disturbance models will be studied. In addition, robust control approaches will be investigated. While TP1 focuses on the corresponding system-specific problem formulation, in particular on the adequate description of the uncertainty, the developed algorithms will be expanded in TP2 to meet the high computational demands despite the increasing operations resulting from the robust approach.Finally, the developed control algorithms are validated on the engine test bed together with TP3. Decisive criteria are the covered operating map, in which stable operation can be realized while taking into account transient load and speed profiles as well as the potential for reducing emissions and increasing efficiency.