Applications of Artificial Intelligence Algorithms for Engine-Out NOx Emissions

Date
2023
Authors
Garcia Hernandez, Stephanie B.
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract

The transportation sector is the major producer of greenhouse gas emissions derived from the burning of fossil fuels when energy is released from the chemical process of fuel and air mixture in an internal combustion engine. The reaction between nitrogen and oxygen during elevated temperature combustion forms the chemical pollutant, Nitrogen Oxide (NOx), the leading agent of air pollution. NOx emissions can be controlled and diagnosed through a model by regulating the inputs, engine operating parameters, to yield an output such as NOx. With the successful applications of artificial intelligence (AI) in image processing and text mining, the goal of this study is to explore three different artificial intelligence algorithms including the Multi-Layer Perceptron (MLP), Long Short-Term Memory (LSTM), and Convolutional Neural Network (CNN), to model the heavy-duty (HV) diesel engine and predict engine-out NOx. Inputs to the model include engine speed, engine torque, fuel rail pressure, exhaust gas recirculation (EGR) valve command, and main injection timing. The R2 from three AI models are above 0.90. CNN and MLP can predict NOx emissions within a 10% error band. Specifically, the MLP algorithm demonstrated the best performance with a test MAPE at 8.62% compared with the 15% MAPE from the LSTM algorithm. Additionally, LSTM has the best training accuracy, and the CNN model shows better feature extraction. The combined CNN-SLTM is constructed to further examine the prediction accuracy, leading to an improved performance with 7.13 % MAPE compared against LSTM or CNN algorithms.

Description
This item is available only to currently enrolled UTSA students, faculty or staff. To download, navigate to Log In in the top right-hand corner of this screen, then select Log in with my UTSA ID.
Keywords
Convolutional neural network, Long short term memory, Machine learning, Multi-layer perceptron, artificial intelligence
Citation
Department
Electrical and Computer Engineering