Multi-objective optimization of Fe-based SCR catalyst on the NOx conversion efficiency for a diesel engine based on FGRA-ANN/RF

With the development of industrial level and the increasingly strict emission regulations, selective catalytic reduction (SCR) system is one of the important ways to control NOx emissions from diesel engines. In this study, the effects of intake pressure on pressure and NOx conversion efficiency before and after catalyst were investigated at different loads. In order to improve the SCR efficiency, a multi-objective prediction method based on FGRA-ANN/RF was developed for SCR systems, where the predicted output parameters (the front and back section temperatures of the catalyst, the pressure difference, and the instantaneous catalytic efficiency) and the decision variables (the engine operating parameters). Firstly, the sensitivity analysis of the multi-objective inputs obtained from the ETC condition experiments is performed by the fuzzy grey relation method (FGRA) to select the high sensitivity parameters. Then, the prediction was carried out by artificial neural network (ANN) as well as the improved random forest (RF). The experimental results showed that the evaluation index R2 of RF was generally greater than 0.92 and the improved random forest method had high accuracy and robustness. It is valuable for solving SCR industrial simulation and reducing the hysteresis of ammonia injection response.