Research On Robustness And Algorithm Of Sonar Beamforming

The detection and estimation performance of a sonar array system would degrade whenit works in formidable acoustic conditions or there exist strong interferences. To improve theperformace, theoretical and experimental researches on the robustness of beamformers arecarried out according to the latest trends of beamforming technique and the practical acousticsituations. The robustness mentioned in this dissertion is for the array model errors and themoving strong interferences. The key contributions are:1. For standard Capon beamforming algorithm is sensitive to array model errors, a novelapproach of adptive beamforming based on double factors diagonal loading technology isproposed. This approach combines the idea of ridge estimation in linear regression model andthe idea of general linear combination in covariance matrix estimation, needs no parameterspecified by users, and automatically computes the diagonal loading factors according to thedata received by the array. Numerical simulation results show effectiveness of the proposedapproach and its robustness against the array model errors.2. Lots of robust adaptivie beamforming algorithms have been designed for the arrays ofomnidirectional sensors. Some of them are migrated straightforwardly to the application ofvector-sensor array but only result in suboptimal performance. To resolve this problem, avector-sensor array adaptive beamforming approach based on double norm constraints ispresented. Vector sensor consists of omnidirectional sensor and particle motion sensor whichhave different response and noise characteristics and are inconsistent in sensitivity to arraymodel errors. According to characteristics of vector sensor, the proposed approach specifiesdifferent values of norm constraint for pressure component and particle velocity component tomeet their own need of robustness. It is can be seen from the numerical simulation results thatthe new approach show its robustness against channel amplitude and phase errors, sensorposition error, sensor attitude errors, beam pointing error and covariance matrix error.3. For conventional broadband adaptive beamforming algorithms can only provide sharpnulls in the direction of interferences and are lack of robustness against moving stronginterferences, three broadband beamforming approaches with broad nulls are proposed,namely Mailloux-based approach, Zatman-based approach and Chebyshev-criteron-basedapproach. Unlike the first two approaches that can produce adaptively broad nulls arounddirections of interferences, the third approach needs to know approximate directions ofinterferences. But the beam pattern designed with the third approach has a constant width of main lobe which can avoid the distortion of signals after beamforming. Numerical simulationresults show the effectiveness of these three new approaches and their robustness againstmoving strong interferences.4. Influenced by moving strong interferences, the measurement precision of weak targetsources would decrease in underwater image measurement. A novel approach of near fieldfocused beamforming is proposed. This approach is based on matrix spatial prefilteringtechnology. By using the idea of designing stop-band matrix filter in plane-wave condition,this new approach can set multiple stop-band regions in the scanning area to suppress themoving strong interferences. Comparison of conventional focused beamforming, MVDRfocused beamforming, null-weight focused beamforming and the new approach throughnumerical simulation shows the superior robustness of the new approach against movingstrong coherent interferences.