Neural Adaptive Ignition Control

To be able to meet the demands of low emissions and low fuel consumption of modern combustion engines, new ways have to be found to control the engine efficiently. We measure the pressure in the combustion chamber and analyze this signal with a neural network in order to receive the point of 50% conversion of energy.

Using this on-line computation of the point of 50% conversion of energy, we construct a linear feedback controller and a neural controller for the computation of the optimal ignition time.

Several complex algorithms for the adaptation of the neural networks are employed and compared in this challenging application. All algorithms were implemented in an ECU and applied to the control of a Mercedes-Benz SI engine. Extensive engine tests were carried out and several results are shown in this article.