Determination of Air Fuel Ratio in Automotive Ignition System Using Neural Networks

This paper describes the use of a neural network in a technique suitable for diagnostic or condition monitoring applications in a spark-ignition engine. The method involves the use of voltage and current variations at the spark plug under normal running conditions, together with neural network analysis. Waveforms produced under test using this technique contain information relating to the adjustment of the engine, for example, the airfuel ratio, ignition timing etc. This means of data acquisition is considerably simpler and more robust than comparable techniques. It is shown that neural networks can be trained to perform data abstraction and analysis on these signals, leading to the determination of the airfuel ratio.