Engine Control Improvement through Application of Chaotic Time Series Analysis

Green, J.B., Jr. ; Daw, C.S.

Oak Ridge National Laboratory

关键词: Data Analysis;    Nitrogen Oxides;    Combustion;    Feedback;    Engines;   

DOI  :  10.2172/940373
RP-ID  :  ORNL95-0337
RP-ID  :  DE-AC05-00OR22725
RP-ID  :  940373

美国|英语

来源: UNT Digital Library

【 摘 要 】

The objective of this program was to investigate cyclic variations in spark-ignition (SI) engines under lean fueling conditions and to develop options to reduce emissions of nitrogen oxides (NOx) and particulate matter (PM) in compression-ignition direct-injection (CIDI) engines at high exhaust gas recirculation (EGR) rates. The CIDI activity builds upon an earlier collaboration between ORNL and Ford examining combustion instabilities in SI engines. Under the original CRADA, the principal objective was to understand the fundamental causes of combustion instability in spark-ignition engines operating with lean fueling. The results of this earlier activity demonstrated that such combustion instabilities are dominated by the effects of residual gas remaining in each cylinder from one cycle to the next. A very simple, low-order model was developed that explained the observed combustion instability as a noisy nonlinear dynamical process. The model concept lead to development of a real-time control strategy that could be employed to significantly reduce cyclic variations in real engines using existing sensors and engine control systems. This collaboration led to the issuance of a joint patent for spark-ignition engine control. After a few years, the CRADA was modified to focus more on EGR and CIDI engines. The modified CRADA examined relationships between EGR, combustion, and emissions in CIDI engines. Information from CIDI engine experiments, data analysis, and modeling were employed to identify and characterize new combustion regimes where it is possible to simultaneously achieve significant reductions in NOx and PM emissions. These results were also used to develop an on-line combustion diagnostic (virtual sensor) to make cycle-resolved combustion quality assessments for active feedback control. Extensive experiments on engines at Ford and ORNL led to the development of the virtual sensor concept that may be able to detect simultaneous reductions in NOx and PM emissions under low temperature combustion (LTC) regimes. An invention disclosure was submitted to ORNL for the virtual sensor under the CRADA. Industrial in-kind support was available throughout the project period. Review of the research results were carried out on a regular basis (annual reports and meetings) followed by suggestions for improvement in ongoing work and direction for future work. A significant portion of the industrial support was in the form of experimentation, data analysis, data exchange, and technical consultation.