During the last years, a number of techniques aimed at the experimental identification of the knocking onset in Spark-Ignition (SI) Internal Combustion Engines have been proposed. Besides the traditional procedures based on the processing of in-cylinder pressure data in the frequency domain, in the present paper two innovative methods are developed and compared.
The first one is based on the use of statistical analysis by applying an Auto Regressive Moving Average (ARMA) technique, coupled to a prediction algorithm. It is shown that such parametric model, applied to the instantaneous in-cylinder pressure measurements, is highly sensitive to knock occurrence and is able to identify soft or heavy knock presence in different engine operating conditions.
An alternative, more expensive procedure is developed and compared to the previous one. The latter is based on the solution of a kinetic scheme in the end-gas zone, whose thermodynamic conditions are reconstructed by means of a two-zone inverse heat release analysis.
Trains of consecutive experimental pressure cycles are acquired on a “downsized” turbocharged SI engine at full load and for different engine speeds. The above data are processed by the two techniques, and knock occurrence and intensity is estimated through suitably defined indices.
The presented results demonstrate that the proposed methods give similar, although not coincident, results. While the kinetic procedure is able to furnish a more detailed insight of the thermo-kinetic conditions inside the cylinder, the ARMA technique is indeed characterized by a lower computational effort. In addition it may be applied to vibrational signals acquired by low-cost accelerometers, too. For this reason, it can be more easily implemented within a on board real time control system, aiming to adjust the spark advance and avoid abnormal combustion phenomena.