Multivariate Knock Detection for Development and Production Applications

Abstract

Combustion knock caused by end gas autoignition continues to be a limiting factor in the performance of automotive internal combustion engines. As such, the availability of efficient knock detection methods is a prime requirement for the optimization of engine mapping and control. Current production knock control systems are based on the measurement of mechanical vibration induced by the acoustic resonance excited in the combustion chamber during autoignition. These vibrations are measured using accelerometers on the engine block. Conversely, knock detection in the laboratory environment during engine development or calibration generally involves either acoustic methods or acquisition of in-cylinder pressure. The purpose of this study is to develop an improved multi-transducer vibration-based knock detection method with applications in engine development and production. The possibility of replacing the pressure-based detection methods in the laboratory environment presents many advantages relating to cost and efficiency. Moreover, the economy of a vibration-based system coupled with improved correlation to laboratory methods represents great potential for performance improvements if applied to production applications.