Reinforcement-learning based output-feedback controller for nonlinear discrete-time system with application to spark ignition engines operating lean and EGR

Author

Peter Shih

Keywords and Phrases

Exhaust gas recirculation (EGR)

Abstract

"A spark ignition (SI) engine can be described by non-strict feedback nonlinear discrete-time system with the output dependent upon on the states in a nonlinear manner. The controller developed in this thesis utilizes the inherent universal approximation property of neural networks (NN) to simplify the design process and solve the non-causality problem inherent with traditional designs. It also exploits a long-term performance index called the strategic utility function to minimize and assist in updating of the NN weights; therefore, an optimal controller can be realized. Finally, through Lyapunov equations, the controller guarantees stability"--Abstract, page iv.

Advisor(s)

Sarangapani, Jagannathan, 1965-

Committee Member(s)

Smith, Scott C.
Drallmeier, J. A.

Department(s)

Electrical and Computer Engineering

Degree Name

M.S. in Computer Engineering

Publisher

University of Missouri--Rolla

Publication Date

Spring 2007

Pagination

xi, 93 pages

Rights

© 2007 Peter Shih, All rights reserved.

Document Type

Thesis - Open Access

File Type

text

Language

English

Subject Headings

Automobiles -- Motors -- Computer control systemsAutomobiles -- Motors -- Exhaust gasNeural networks (Computer science)Reinforcement learning (Machine learning)

Thesis Number

T 9155

Print OCLC #

173691079

Electronic OCLC #

128261909

Recommended Citation

Shih, Peter, "Reinforcement-learning based output-feedback controller for nonlinear discrete-time system with application to spark ignition engines operating lean and EGR" (2007). Masters Theses. 4550.
https://scholarsmine.mst.edu/masters_theses/4550