Engine dynamics identification in the case of non-uniformly sampled data

Irregular sampling occurs in several applications such as geophysics, laser Doppler anemometry or engine control. In this paper a new parametric spectral estimation method is addressed. It is based on a continuous time ARMA model of the signal. A specific inversion criterion is defined to estimate spectral parameters. This criterion is based on the difference between two estimators of the correlation function. A specific algorithm has been designed to minimize this criterion. The whole procedure is applied to space-time conversion in the automotive field. Results show good performances even when the mean sampling frequency is below the Nyquist frequency.