| Signal direction of arrival(DOA)technology is a popular research direction in array signal processing.The traditional beam-type DOA estimation method cannot break through the Rayleigh limit.The subspace-type method brings super-resolution,but This type of method has higher requirements for the signal-to-noise ratio and the number of snapshots,and additional constraints need to be imposed in the non-Gaussian case.Aiming at the demand for high-precision DOA estimation in complex underwater acoustic environments,this paper uses the sparseness of the signal and the flexibility of sparse Bayesian modeling to study the DOA estimation method based on sparse Bayesian learning.The main research contents are as follows:First,this thesis studies several traditional DOA estimation methods for far-field narrowband signals,including Conventional Beamforming(CBF)algorithm,Minimum Variance Distortionless Response(MVDR)algorithm and Multiple Signal Classification(MUSIC)algorithm,and then introduces the Sparse Bayesian Learning(SBL)algorithm and the M-SBL algorithm.On this basis,the variational SBL algorithm is introduced and used for DOA estimation.And the simulation comparisons of various algorithms are performed under the background of Gaussian noise.The results show that the estimation accuracy of sparse Bayesian learning algorithms is significantly higher than that of traditional algorithms,the requirement for the number of snapshots is low,and it can handle coherent signals.Subsequently,using the modeling flexibility of the variational SBL theory,the DOA estimation methods in the presence of the amplitude and phase errors of the array,the background noise of the array being non-uniform,and the presence of impulse interference were studied respectively,and compared with the existing methods.When amplitude and phase errors exist,the proposed method does not need auxiliary correction sources,and the estimation accuracy is higher than that of traditional correction methods.When the background noise of the array is non-uniform noise ,the method in this paper requires less snapshots than the MUSIC-based noise covariance matrix estimation method,and has a higher angular resolution,and has a stronger ability to suppress non-uniform noise .When the array has impulse noise interference,the impulse noise is regarded as the student’s t distribution for modeling.Compared with the Variable Step Size Least Mean Square(VSSLMS)algorithm,the method in this paper has better performance,and this method is also effective in the Gaussian noise environment.Finally,in view of the simultaneous existence of the array amplitude and phase error and the non-uniform background noise,the joint estimation method based on the variational SBL algorithm is studied,and the effectiveness of the method is verified by computer simulation,and then the performance of the narrowband signal algorithm is weakened when processing real data.The broadband signal processing method is studied,and the sea trial data is processed and verified.The results show that the estimation accuracy of the proposed method is significantly higher than that of the traditional DOA estimation method,and it can effectively detect the ship’s azimuth. |
