Implementation of an embedded neural network-based output-feedback controller for spark ignition engines

Author

Jonathan B. Vance

Keywords and Phrases

Exhaust gas recirculation

Abstract

"A neural-network-based output-feedback controller is employed on a spark ignition engine to control fuel input during lean operation for fuel-air equivalence ratios less than 1.0 and in the presence of exhaust gas recirculation (EGR). At lean operating conditions, the engine exhibits significant cycle-to-cycle bifurcation of heat release -- a characteristic of instability. The controller reduces the cyclic dispersion of heat release by minimizing the error between target heat release and measured engine heat release for every engine cycle in real time. Consequences of operating a spark-ignition engine under lean conditions are reduction of harmful NOx emissions and an improvement in fuel economy.

When the controller is operating the engine, experimental results show that lean operation at decreased equivalence ratios returns similar results to lean engine operation from increased levels of EGR. The cyclic dispersion decreases when output-feedback control is applied to the fuel input. Moreover, controller operation at a lean equivalence ratio of 0.77 yields a 98.6% drop in NOx emissions from stoichiometric operation. Similarly, an 80% NOx drop is seen when EGR is increased to 10%. Applying control at an equivalence ratio of 0.77 caused cyclic dispersion to decrease 64.6% and unburned hydrocarbons to decrease 26.8%. For all observed operating set points, the output- feedback controller increases the equivalence ratio by about 5% which includes measurement errors and target operating point specification errors. The error in equivalence ratio is caused by limitations of fuel as a control input"--Abstract, page iii.

Advisor(s)

Sarangapani, Jagannathan, 1965-

Committee Member(s)

Drallmeier, J. A.
Smith, Scott C.

Department(s)

Electrical and Computer Engineering

Degree Name

M.S. in Computer Engineering

Publisher

University of Missouri--Rolla

Publication Date

Fall 2005

Pagination

ix, 93 pages

Note about bibliography

Includes bibliographical references (page 92).

Rights

© 2005 Jonathan Blake Vance, All rights reserved.

Document Type

Thesis - Restricted Access

File Type

text

Language

English

Subject Headings

Neural networks (Computer science)Spark ignition engines -- Exhaust gasNitrogen oxides

Thesis Number

T 8895

Print OCLC #

71003887

Recommended Citation

Vance, Jonathan B., "Implementation of an embedded neural network-based output-feedback controller for spark ignition engines" (2005). Masters Theses. 5832.
https://scholarsmine.mst.edu/masters_theses/5832