Maximum Likelihood Estimation of Propeller Noise Modulation Characteristics

Propeller cavitation noise in the high frequency regime is broadband and yet audibly rhythmic. The study of these processes are important because the shaft and blade rates, as well as other identifying features of the ship, can in principle be inferred from propeller noise modulation characteristics. Conventional wisdom has been that propeller noise is an amplitude modulation of a wide sense stationary random process carrier. We show here that this assumption is wrong and that propeller noise actually often has a time-dependent spectral shape in addition to amplitude modulation characteristics. Motivated by this, we model the ship’s propeller noise as a general periodicallycorrelated (PC) random process which allows for time-dependent spectral characteristics in addition to amplitude modulation and then derive its maximum likelihood estimator under the weak signal assumption. The new estimator is demonstrated on actual merchant ship noise data.