Improved Knock Detection Method Based on New Time-Frequency Analysis In Spark Ignition Turbocharged Engine

Premature combustion that affects outputs, thermal efficiencies and lifetimes of internal combustion engine is called “knock effect”. However knock signal detection based on acoustic sensor is a challenging task due to existing of noise in the same frequency spectrum. Experimental results revealed that vibration signals, generated from knock, has certain frequencies related to vibration resonance modes of the combustion chamber. In this article, a new method for knock detection based on resonance frequency analysis of the knock sensor signal is introduced. More specifically at higher engine speed, where there is additional excitation of resonance frequencies, continuous wavelet transform has been proposed as an effective and applicative tool for knock detection and a formula for knock detection threshold based on this method is suggested. Measurement results demonstrate that this technique provide 15% higher accuracy in knock detection comparing to conventional method.