Prediction Model for Transient NOx Emission of Diesel Engine Based on CNN-LSTM Network

In order to address the challenge of accurately predicting nitrogen oxide (NOx) emission from diesel engines in transient operation using traditional neural network models, this study proposes a NOx forecasting model based on hybrid architecture combining convolutional (CNN) and long short-term memory (LSTM) network. The objective is enhance calibration efficiency reduce engine emissions. proposed utilizes data collected under thermal cycle according world harmonized (WHTC) test standard for training verifying prediction model. CNN employed extract features data, while LSTM networks are used fit resulting precise emissions engines. Experimental verification was conducted results demonstrate that fitting coefficient (R2) CNN-LSTM 0.977 with root mean square error 33.495. Compared predictions made by single network, predictions, back-propagation (BP) (RMSE) decreases 35.6%, 50.8%, 62.9%, respectively, degree R2 increases 2.5%, 4.4%, 6.6%. These has higher accuracy, good convergence, robustness.