Abstract: The traditional Active Noise (ANC) system does not work effectively in the environment where the noise is mixed with the useful signal. An active noise cancellation system applied on the passive sonar platform of warships is designed to solve the problem said above, which can enhance the role range of the passive sonar. This ANC system uses an FX structure based on wavelet packet decomposition. Because of the orthogonal characteristics of wavelet packet decomposition, the correlation between every two weights is reduced, and the distribution of reference-input matrix’s eigenvalues is also improved. The working method of the system’s filter-group ensures that the noise in different frequency bands can be distinguished and cancelled respectively. So the system has the effect that is the same as that of the multi-frequency-band system. Combined with the LMS algorithm, this system has the flexible whole-band distinguish ability in both the time-domain and the frequency domain and a faster convergence rate. Simulations are done based on practical sea-trial data. In the condition of smaller distortion of signals, a 12.2dB noise reduction is achieved over the whole-band. It is indicated that the system designed in this paper has advantages in the complex underwater noise field. This system enhances the noise controlling capability. So it has a high practical value.

Key words: acoustic, adaptive active noise cancellation, wavelet, Filtered-XLMS, self-noise