Minimum dispersion-based cyclic reconstruction beamforming for propeller noise signal

The minimum dispersion distortionless response (MDDR) beamformer outperforms the minimum variance distortionless response (MVDR) beamformer for non-Gaussian signals, but suffers from ambiguous cost function selection and degraded performance against close-angle non-planar wave interferences. To address these issues, based on the super-Gaussian characteristics and cyclostationarity of propeller noise, this paper presents a minimum dispersion-based cyclic reconstruction (MDCR) beamforming algorithm. This algorithm optimizes the design of the cost function by clarifying the quantitative relationship between the norm of the cost function in the traditional minimum dispersion algorithm and the physical parameters of propeller noise, such as the shaft frequency and bubble burst time. Simultaneously, the algorithm introduces the circular spatial spectrum reconstruction steering vectors based on the shaft frequency of the target or interference, thereby enhancing its adaptability to strong interferences and near-field targets. The simulation experiment results indicate that, in comparison with traditional methods, the output signal-to-noise ratio (SNR) of the proposed method increases by 1.0 dB when processing near-field propeller noise; when there are interference signals at close distances, through optimizing the norm setting and steering vector reconstruction, the SNR increase reaches 4.2 dB. Additionally, the verification of sea trial data demonstrates that this algorithm possesses considerable application potential in a strong interference acoustic environment.