Knock detection based on MAPO analysis, AR model and discrete wavelet transformation applied to the in-cylinder pressure data:results and comparison

The easiest way to identify knock conditions during the operation of a SI engine is represented by the knowledge of the in-cylinder pressure. Traditional techniques like MAPO (Maximum Amplitude Pressure Oscillation) based method rely on the frequency domain processing of the pressure data. This technique may present uncertainties due to the correct specification of some model parameters, like the band-pass frequency range and the crank angle window of interest. In this paper two innovative techniques for knock detection, which make use of the in-cylinder pressure, are explained in detail, and the results are compared with those coming from the MAPO method. The first procedure is based on the use of statistical analysis by applying an Auto Regressive (AR) technique, while the second technique makes use of the Discrete Wavelet Transform (DWT). The data useful for the analysis have been acquired on a high compression ratio four cylinder, spark ignition engine. Results demonstrate that the analyzed methods give quite similar outcomes but they also highlight that AR and DWT techniques present an higher sensitivity for soft knock detection. Moreover, both AR model and DWT based method may be applied to the engine block vibrational signals, acquired by low-cost accelerometers. Therefore, they can be implemented within modern on-board control units, in order to execute a real time knock detection allowing to realize more sophisticated spark advance control strategies.