Emissions Modeling of a Light-Duty Diesel Engine for Model-Based Control Design Using Multi-Layer Perceptron Neural Networks

2017-01-0601

03/28/2017

Features
Event
WCX™ 17: SAE World Congress Experience
Authors Abstract
Content
The development of advanced model-based engine control strategies, such as economic model predictive control (eMPC) for diesel engine fuel economy and emission optimization, requires accurate and low-complexity models for controller design validation. This paper presents the NOx and smoke emissions modeling of a light duty diesel engine equipped with a variable geometry turbocharger (VGT) and a high pressure exhaust gas recirculation (EGR) system. Such emission models can be integrated with an existing air path model into a complete engine mean value model (MVM), which can predict engine behavior at different operating conditions for controller design and validation before physical engine tests. The NOx and smoke emission models adopt an artificial neural network (ANN) approach with Multi-Layer Perceptron (MLP) architectures. The networks are trained and validated using experimental data collected from engine bench tests. Model inputs (including input delays) are selected based on physics-based analyses supplemented with data-driven cross-covariance studies. Special care is taken during the training process to avoid overfitting and ensure strong generalization performance. Various neural network architectures, including static networks, dynamic networks, and classifiers, are compared in terms of model complexity and accuracy. Simulation results indicate that MLP networks are capable of capturing the highly nonlinear engine NOx and smoke emissions at both steady state and transient conditions.
Meta TagsDetails
DOI
https://doi.org/10.4271/2017-01-0601
Pages
9
Citation
Li, H., Butts, K., Zaseck, K., Liao-McPherson, D. et al., "Emissions Modeling of a Light-Duty Diesel Engine for Model-Based Control Design Using Multi-Layer Perceptron Neural Networks," SAE Technical Paper 2017-01-0601, 2017, https://doi.org/10.4271/2017-01-0601.
Additional Details
Publisher
Published
Mar 28, 2017
Product Code
2017-01-0601
Content Type
Technical Paper
Language
English