Model Identification for Optimal Diesel Emissions Control

Conference · Thu Jun 20 00:00:00 EDT 2013

OSTI ID:1089079

Stevens, Andrew J.; Sun, Yannan; Song, Xiaobo; Parker, Gordon

In this paper we develop a model based con- troller for diesel emission reduction using system identification methods. Specifically, our method minimizes the downstream readings from a production NOx sensor while injecting a minimal amount of urea upstream. Based on the linear quadratic estimator we derive the closed form solution to a cost function that accounts for the case some of the system inputs are not controllable. Our cost function can also be tuned to trade-off between input usage and output optimization. Our approach performs better than a production controller in simulation. Our NOx conversion efficiency was 92.7% while the production controller achieved 92.4%. For NH3 conversion, our efficiency was 98.7% compared to 88.5% for the production controller.

• Research Organization: Pacific Northwest National Lab. (PNNL), Richland, WA (United States)
• Sponsoring Organization: USDOE
• DOE Contract Number: AC05-76RL01830
• OSTI ID: 1089079
• Report Number(s): PNNL-SA-94339; 830403000
• Resource Relation: Conference: Proceedings of the 30th International Conference on Machine Learning (ICML), June 16-21 2013, Atlanta, Georgia, 28
• Country of Publication: United States
• Language: English