This paper proposes a discrete-time preview-feedback controller capable of reducing nitrogen oxide (NO_x) emissions with affordably low ammonia (NH₃) slip downstream of a catalyst for selective catalytic reduction (SCR) systems. Because of the high NO_x reduction rate, SCR systems are increasingly adopted in diesel engines. However, slow dynamics of the SCR system along with a time delay in the sensor result in a challenging control problem. To overcome the control challenges, the preview-feedback controller is proposed under the assumption that the NO_x disturbance is previewed. To design the preview-feedback H_∞ controller, a discrete-time generalized plant is derived by combining the linearized model of SCR system, the model for previewing disturbance signal, and the weighting function. Based on experimentally obtained data, the simulations show that the proposed controller can effectively reduce 34 % of NO_x emissions and 6 % of NH₃ slip more, compared to a non-preview controller. The robustness is then demonstrated using the uncertainties ranging from 10 to 40 %. This result shows that uncertainties of up to 30 % do not negatively affect the performance. Additionally, the result for the computational requirement shows that the proposed controller yields better performance with lower computation time than a model predictive controller does.

本文提出了一种离散时间预览反馈控制器，该控制器能够在选择性催化还原 (SCR) 系统的催化剂下游以经济实惠的低氨 ( NH ₃ ) 泄漏减少氮氧化物 ( NO _x ) 排放。由于高NO _x还原率，SCR 系统越来越多地用于柴油发动机。然而，SCR 系统的缓慢动态以及传感器中的时间延迟导致了具有挑战性的控制问题。为了克服控制挑战，在预测NO _x扰动的假设下提出了预览反馈控制器。设计预览反馈H _∞控制器结合SCR系统的线性化模型、扰动信号预览模型和加权函数，推导出离散时间广义对象。根据实验获得的数据，所述仿真结果表明，所提出的控制器能有效地降低了34％NO _X排放量和的6％NH ₃滑移更多，相对于非预览控制器。然后使用 10% 到 40% 的不确定性来证明稳健性。该结果表明，高达 30% 的不确定性不会对性能产生负面影响。此外，计算要求的结果表明，与模型预测控制器相比，所提出的控制器以更少的计算时间产生更好的性能。