柴油机SCR载体SOF沉积量估算模型与参数辨识

针对柴油机选择性催化还原(Selective Catalytic Reduction,SCR)载体由于可溶性有机物(Soluble Organic Fraction,SOF)沉积导致SCR的NO_x转化效率低的问题,该研究首先建立了SOF沉积量估算模型,并进行SOF原始排放与SCR载体的SOF捕集效率试验研究;然后利用Matlab/Simulink软件工具建立SOF沉积量估算模型,包括SOF原始排放模块、SCR载体对SOF捕集模块、SOF热解模块;最后基于多目标遗传算法,进行SOF瞬态修正脉谱优化和热解参数辨识,并探索惩罚函数的应用规律,使得4组SOF低温沉积量平均估算误差为2.42%,12组高温热解平均估算误差为4.03%。该研究可为解决柴油机SCR载体由于SOF沉积导致NO_x转化效率低的问题提供技术参考。

Estimation model and parameter identification of SOF deposition on SCR carrier of diesel engines

Abstract: This study aims to improve the low NOx conversion efficiency of the selective catalytic reduction (SCR) system caused by the soluble organic fraction (SOF) deposition of SCR carrier in a diesel engine. A sediment quantity model of SOF deposition was proposed to optimize key parameters using a multi-objective genetic algorithm (GA). Firstly, a road spectrum test at low temperatures was conducted to collect the data of SOF deposition amount in four groups. A SOF high-temperature pyrolysis was carried out to obtain 10 groups of experimental data at a steady-state temperature. Secondly, a model of SOF deposition was established using Matlab/Simulink tools, including SOF raw emission, the SCR carrier of SOF capture, and SOF pyrolysis module. Two ways were selected to calculate the SOF raw emission. One was the theoretical estimation using an approximate linear relationship of SOF with the gaseous unburned hydrocarbons (HC). Another was the direct measurement of SOF raw emission, where the excess air coefficient was used to correct the transient SOF raw emission. The capture efficiency of SOF by SCR carrier was evaluated via mapping the upstream discharge and exhaust gas flow of SCR. A correction was also introduced using the deposition amount of SOF. The SOF pyrolysis was prepared under the component analysis and chemical reaction kinetics model of SOF. Three stages were divided in a pyrolysis process of SOF, including the short, medium, and long chain. The key parameters of SOF pyrolysis were determined, such as the transient correction MAP, activation energy of three stages, pre-exponential factor, and mass proportion coefficient. Thirdly, various multi-objective GAs were evaluated prior to optimization. An interactive adaptive-weight GA (i-awGA) was selected to optimize the key parameters considering both efficiency and accuracy, whereas, a non-dominated sorting GA II (nsGA II) was used to identify the optimal solution, and a strength Pareto evolutionary algorithm (spEA) was utilized to generate the penalty function. Finally, a multi-objective GA optimization was performed on the transient correction MAP and three groups of pyrolysis parameters. In MAP optimization, the number, range, and distribution of MAP points acted by each gene were calculated using the type of MAP, combined with the basic value of MAP and the number of genes. In optimization of pyrolysis parameters, pre-exponential factors were calculated using the activation energy and compensation effect for the physical significance of the model. The average error of low temperature deposition of 4 groups of SOF reached 2.42%, the average error of high temperature pyrolysis of 12 groups reached 4.03%. Specifically, the largest average error of 3.04% was obtained for the low-temperature path deposition in one group of SOF, whereas, the largest average error of 5.41% was for the steady-temperature pyrolysis in two groups. It demonstrates that the proposed model of SOF deposition and the GA optimization was well suitable for the engineering application.