基于偏微分建模及滚动优化的柴油机urea-SCR系统控制研究

Urea selective catalytic reduction (urea-SCR) is an important emissions aftertreatment technology for diesel engine. It is a difficult problem in the field of engineering that how to adjust the amount of urea injection through the electronic control system so that contradictory NOx and NH3 emissions can meet the emission standards. The system is a typical distributed parameter system with multi matter unevenly propagated along the axial direction of the catalyst and set off complex chemical reactions. The system has infinite dimensional characteristics and large hysteresis and uncertainty, the existing ordinary differential equations and control methods are difficult to achieve the desired control performance. Therefore, this moving horizon optimization control research based on partial differential equation (PDE) model will be conducted by means of experimental analysis. 1. Starting from the bench analysis, influence of different initial boundary conditions on emission are analyzed, and sensitive factor of modeling and optimization control are determined; 2. A moving horizon optimization control method based on PDE model is proposed according to multi-objective optimization problems with emission constraints; 3. An integrated estimation method with the overall stability and robustness is proposed according to round-circle coupled estimation problem of unmeasured states; 4. Some theoretical results are implemented using FPGA on experimental test beds considering the problem of real-time and miniaturization. A new algorithm is proposed by combining PDE modeling and moving horizon optimization based on the study of urea-SCR system emission control. The project has great practical significance in engineering application and theoretical research.

尿素选择性催化还原（urea-SCR）是一项重要的柴油机排放后处理技术。如何通过电控调节尿素喷射量，使得互相矛盾的NOx和NH3排放都能满足排放标准，是工程领域的一个难题。该系统是一个典型的多物质沿催化器轴向不均匀传播、并发生复杂化学反应的分布参数系统。该系统具有无限维特性以及较大的滞后和不确定性，现有的常微分方程及控制方法很难达到理想控制效果。因此，拟在实验分析基础上进行基于偏微分方程（PDE）模型的滚动优化控制研究。1、台架试验分析各边界条件对排放的影响，确定建模及优化控制敏感因素；2、针对排放约束的多目标多变量优化问题，建立PDE模型并提出滚动优化控制方法；3、针对不可测状态的循环耦合估计问题，提出具有稳定性和鲁棒性的一体化估计方法；4、考虑实时性和小型化问题，基于FPGA实现高效低成本方法。将PDE建模与滚动优化相结合提出排放控制新方法，在工程应用和理论研究中都有较大的实际意义。

在本项目的资助下，围绕着urea-SCR系统控制问题，主要开展了以下几个方面的工作。1、台架试验分析各边界条件对排放的影响，确定建模及优化控制敏感因素；2、针对排放约束的多目标多变量优化问题，建立PDE模型并提出滚动优化控制方法；3、在ISS稳定性理论框架下设计了具有整体稳定性的一体化氨覆盖率观测器和控制器。在尿素转化为氨不确定性的干扰条件下，讨论了观测器和跟踪控制器的整体稳定性；4、本项目设计了一种面向节能减排的车辆车速、发动机以及尿素SCR后处理系统序列优化控制方法，该方法能够降低5.3%的燃油消耗，并且在加速阶段能够降低11.56%的PM排放。5、针对柴油混合动力车辆能量分配和后处理系统热管理的多目标多约束的节能减排问题，本项目提出了一种分层级的能量管理和热管理模型预测控制方法。6、针对某型号SCR系统实际的排放和尿素成本需求，本项目从模型参数辨识、状态估计、排放优化控制，到autobox中控制算法实时优化等方面，以模型预测优化控制技术为核心解决了企业提出的实际问题。