Combustion Quality Monitoring Using Neural Network Analysis of Ignition Spark Voltage Vectors

In recent years, the gasoline-fuelled internal combustion (IC) engine has been the subject of a constant drive towards improved fuel economy and increasingly stringent legislation concerning emissions. Both of these factors depend significantly on the quality of combustion that can be obtained, under a multitude of different engine running conditions. Misfire can cause serious damage to a catalytic converter and every occurrence should be logged together with the identity of the offending cylinder. No proposed method of misfire detection has been without shortcomings; often simply the high cost. The spark plug has been used successfully as both combustion initiator and sensor in several IC engine monitoring applications; it is a rugged and cost-effective sensor. In this work a virtual sensor is demonstrated, using the spark plug and a custom neural network to estimate the combustion quality on the basis of relevant data extracted from ignition spark voltage vectors. The virtual sensor was found to exhibit good off-line correlation of combustion quality with its output signals. This type of virtual combustion quality sensor has good potential for providing a relatively simple and cost-effective solution by implementation in software using existing hardware. Dr. S. D. Walters has a Ph.D, an MEng and a BEng(Hons) from the University of Brighton. He is an Associate Member of the Institution of Electrical Engineers and is employed as a Research Fellow at the University of Brighton. He has worked on automotive engine and ignition system component monitoring and fault diagnosis research projects since 1993. His research interests are varied, encompassing condition monitoring, intelligent systems, automotive engineering, and high voltage engineering applications. Dr. R. J. Howlett has a Ph.D from the University of Brighton, an MPhil from the University of Sussex and a BSc(Hons) from Portsmouth Polytechnic. He is a Member of the British Computer Society and a Chartered Engineer. He is Director of the Brighton TCS Centre which runs technology and business improvement projects in small companies. His research interests centre on the industrial applications of intelligent techniques.