Research of Gasoline Engine Ignition Control Based on Neural Network

Abstract:

In order to realize the precise ignition control of gasoline engine, an ignition advance angle BP (Back Propagation) neural network model is built. The improved LM (Levenberg-Marquardt) learning algorithm is used in the model to increase the neural network performance. The neural network model is trained and tested by matlab program. For a variety of inputs, the trained ignition advance angle neural network can carry out correct outputs. Compared with the experimental ignition advance angle, the maximum error of the neural network ignition advance angle is less than 5%. Compared with the experimental map method, the ignition advance angle neural network has the advantage of online modifying the value of ignition advance angle, so it can make the gasoline engine acquire the best ignition advance angle on various working conditions. The results show that the ignition advance angle neural network model established in this paper is effective and accurate. The performance of gasoline engine can be improved ultimately.

• Periodical: Materials Science Forum (Volumes 626-627)
• Pages: 501-504
• DOI: https://doi.org/10.4028/www.scientific.net/MSF.626-627.501
• Online since: August 2009
• Authors: Y.Y. Wang, Jian Guo Yang, B.Y. Song
• Keywords: Gasoline Engine, Ignition Control, Neural Network (NN)

References

[1] Paljoo Yoon: SAE 2000 World Congress(Detroit, America, March 6, 2000).

[2] Cakar T, Cil I: International Journal of Computer Integrated Manufacturing Vol. 17(3) (2004), pp.195-211.

[3] L.Q. Han: The Tutorial of Artificial �eural �etworks(Beijing University of Posts and Telecommunications Press, China 2006)(in chinese).