Approximate dynamic programming solution for the optimal nitrogen oxides/particulate matter trade-off control of a WAPS engine

• Published in: Advances in mechanical engineering Vol. 10; no. 2
• Main Authors: Huang, Zhijian, Zheng, Huan, Chen, Wentao, Zhang, Qin, Wu, Qili, Tan, Qinmin, Yang, Zhiyuan
• Format: Journal Article
• Language: English
• Published: London, England SAGE Publications 01.02.2018; Sage Publications Ltd; SAGE Publishing
• Subjects: Accumulators; Authorship; Carbon; Controllers; Convergence; Diesel engines; Dynamic programming; Economic models; Emission analysis; Emissions control; Engine control; Neural networks; Nitrogen oxides; Optimal control; Particulate emissions; Tradeoffs; Transient performance; China; particulate matter; optimal; nitrogen oxides; WAPS engine; emission control; trade-off; Approximate dynamic programming
• ISSN: 1687-8132; 1687-8140
• DOI: 10.1177/1687814017751781

Approximate dynamic programming is an effective optimal control method. This article researches a data-driven approximate dynamic programming. The method is extended to a nonlinear multi-input multi-output form. Using the data from a unique 4JB1-T weifu accumulator pump system (WAPS) engine, the developed approximate dynamic programming controller is trained to achieve its optimal trade-off emission control between the nitrogen oxides and particulate matter. The convergent proof of this method is given. The second-order training algorithm is introduced to promote the robust and convergent performance. The control objective is to let the WAPS engine pass the China State-IV emission test under the New European Drive Cycle. The bench test shows that an excellent control transient performance and significant promotion have been achieved. This article presents a new approach for the engine control and calibration. In addition, it also adds another dimension to the existing literature on the data-driven nonlinear multi-input multi-output trade-off emission control of the WAPS engine.