Neural combustion control

Abstract

Public as well as legislation request, that future combustion engines have to be lower in fuel consumption and better in terms of emissions. These requirements can only be fulfilled with new and intelligent control strategies, like neural networks. In this paper we present the basics of future combustion control systems before giving a short introduction to recurrent neural networks for the identification and control of nonlinear dynamic systems. Several learning algorithms, like Real Time Recurrent Learning with static and dynamic derivatives and the advanced Kalman Filter algorithm will be compared in the application in new engine management systems.