Ignition Timing Map Calibration Based on Nonlinear Dynamic System Identification Using NARX Neural Network
Apstrakt
The main topic of this paper is presentation of methodology for process identification and mathematical modeling of a nonlinear dynamic system, such as an IC engine, based on the experimental data acquired during base engine calibration in terms of ignition timing. With the introduction of certain assumption, mathematical model generated in this way could be used for verification of potentially optimal look-up tables and for look-up tables smoothing. For this type of time-series modeling, nonlinear autoregressive network with exogenous inputs will be used and full-factorial sweep of limited set of neural network parameters will be analyzed. Guidelines for mathematical model formation, verification and idea of stationary-based engine calibration will be briefly outlined. Comparison between measured and modeled engine torque will be shown alongside with instructions for further research on this topic.
Ključne reči:
Dynamic testing / ECU Calibration / Engine testing / Neural network / Spark timingIzvor:
18th Symposium on Thermal Science and Engineering of Serbia, 2017, 685-693Izdavač:
- University of Niš, Faculty of Mechanical Engineering in Niš
Finansiranje / projekti:
- Istraživanje i razvoj alternativnih pogonskih sistema i goriva za gradske autobuse i komunalna vozila radi poboljšanja energetske efikasnosti i ekoloških karakteristika (RS-MESTD-Technological Development (TD or TR)-35042)
Kolekcije
Institucija/grupa
Mašinski fakultetTY - CONF AU - Mrđa, Predrag D. AU - Miljić, Nenad AU - Popović, Slobodan AU - Kitanović, Marko PY - 2017 UR - https://machinery.mas.bg.ac.rs/handle/123456789/4835 AB - The main topic of this paper is presentation of methodology for process identification and mathematical modeling of a nonlinear dynamic system, such as an IC engine, based on the experimental data acquired during base engine calibration in terms of ignition timing. With the introduction of certain assumption, mathematical model generated in this way could be used for verification of potentially optimal look-up tables and for look-up tables smoothing. For this type of time-series modeling, nonlinear autoregressive network with exogenous inputs will be used and full-factorial sweep of limited set of neural network parameters will be analyzed. Guidelines for mathematical model formation, verification and idea of stationary-based engine calibration will be briefly outlined. Comparison between measured and modeled engine torque will be shown alongside with instructions for further research on this topic. PB - University of Niš, Faculty of Mechanical Engineering in Niš C3 - 18th Symposium on Thermal Science and Engineering of Serbia T1 - Ignition Timing Map Calibration Based on Nonlinear Dynamic System Identification Using NARX Neural Network EP - 693 SP - 685 UR - https://hdl.handle.net/21.15107/rcub_machinery_4835 ER -
@conference{ author = "Mrđa, Predrag D. and Miljić, Nenad and Popović, Slobodan and Kitanović, Marko", year = "2017", abstract = "The main topic of this paper is presentation of methodology for process identification and mathematical modeling of a nonlinear dynamic system, such as an IC engine, based on the experimental data acquired during base engine calibration in terms of ignition timing. With the introduction of certain assumption, mathematical model generated in this way could be used for verification of potentially optimal look-up tables and for look-up tables smoothing. For this type of time-series modeling, nonlinear autoregressive network with exogenous inputs will be used and full-factorial sweep of limited set of neural network parameters will be analyzed. Guidelines for mathematical model formation, verification and idea of stationary-based engine calibration will be briefly outlined. Comparison between measured and modeled engine torque will be shown alongside with instructions for further research on this topic.", publisher = "University of Niš, Faculty of Mechanical Engineering in Niš", journal = "18th Symposium on Thermal Science and Engineering of Serbia", title = "Ignition Timing Map Calibration Based on Nonlinear Dynamic System Identification Using NARX Neural Network", pages = "693-685", url = "https://hdl.handle.net/21.15107/rcub_machinery_4835" }
Mrđa, P. D., Miljić, N., Popović, S.,& Kitanović, M.. (2017). Ignition Timing Map Calibration Based on Nonlinear Dynamic System Identification Using NARX Neural Network. in 18th Symposium on Thermal Science and Engineering of Serbia University of Niš, Faculty of Mechanical Engineering in Niš., 685-693. https://hdl.handle.net/21.15107/rcub_machinery_4835
Mrđa PD, Miljić N, Popović S, Kitanović M. Ignition Timing Map Calibration Based on Nonlinear Dynamic System Identification Using NARX Neural Network. in 18th Symposium on Thermal Science and Engineering of Serbia. 2017;:685-693. https://hdl.handle.net/21.15107/rcub_machinery_4835 .
Mrđa, Predrag D., Miljić, Nenad, Popović, Slobodan, Kitanović, Marko, "Ignition Timing Map Calibration Based on Nonlinear Dynamic System Identification Using NARX Neural Network" in 18th Symposium on Thermal Science and Engineering of Serbia (2017):685-693, https://hdl.handle.net/21.15107/rcub_machinery_4835 .