Research On Modeling And Analysis Of SCR System For Marine Diesel Transient Working Conditions

With the increasingly stringent international and domestic regulations restricting the emission of nitrogen oxides?NOx?from ships,selective catalytic reduction?SCR?technology can effectively reduce the emission of NOx from ships.In emission control zones,such as coastal areas,especially in ports,engines are usually in transient conditions.Accurate establishment of the transient model of SCR system is helpful to improve the accuracy of model-based urea injection control strategy.In this paper,the transient characteristics of SCR system are modeled and analyzed.The main work and conclusions are as follows:1)Aiming at the boundary conditions of SCR system under transient conditions,a three-dimensional simulation model of urea decomposition and mixing process is established and analyzed.The simulation results show that the characteristics of urea deposition and mixing of reductants in the SCR bench system under transient conditions do not affect the catalytic reaction process of the SCR system.2)The experiment of SCR system has been designed in transient condition,the one-dimensional SCR heat transfer and catalytic reaction model has been trained and tested based on boost.In transient condition,SCR heat transfer model can accurately reflect the heat transfer characteristics of SCR system,but the output error of SCR catalytic reaction model is large.3)The SCR catalytic reaction model is established based on BP neural network,and the output performance of the model is improved gradually by optimizing the structure of the neural network,which proves that the single hidden layer network model can reconstruct the traditional SCR model.In addition,compared with the single hidden layer network model,the double hidden layer network model structure can greatly improve the output performance of NOx and NH₃.4)Based on analyses of data and model characteristics,the characteristics of SCR model are comprehensively analyzed,and the consistency evaluation index based on upper and lower bounds of model output error is established.By comparing and analyzing the structure characteristics and output performance of different models,the results show that the output performance of the models with the same complexity is basically the same,and the higher the complexity of the models,the better the output performance.