Application of cyclostationary and time-frequency signal analysis to car engine diagnosis

In modern car engines the angle of ignition has to be controlled adaptively to achieve high efficiency of the engine. This control process is usually based on evaluation of cylinder pressure or structural vibration signals. Due to the nature of the combustion process these signals are highly nonstationary. Therefore signal processing methods well suited to this special kind of signal has to be used. Based on real data experiments we show that a cyclostationary model is appropriate for constant rotation speed of the engine. Use of this model enables us to get a consistent estimate of the Wigner-Ville spectrum of the signals. This time-frequency representation has very good resolution and can give us information about the quality of sensor positions or about important parameters that can be used for adaptive control of the engine.<<ETX>>