Bayesian multichannel tracking of periodic signals: a new way to determine the running speed of mechanical systems

The problem of estimating and tracking the fundamental frequency of a periodic signal is formulated in Bayesian terms and a solution is proposed. For multichannel measurements both simulated and experimental data indicate that the quality of the estimate improves when more data channels are used for estimation. The method is illustrated on a multichannel vibration measurement from the interior of a passenger car. From the measurements the running speed of the engine is successfully estimated during a run-up/down experiment.