Research On Passive Direction Finding Technology For Sources In The Low Frequency Acoustic Vector Field In Shallow Water

Passive direction finding technology for sources in the low frequency acoustic field in shallow water is of critical significance in both of the military and civilian fields.Advancement of the acoustic vector sensor(AVS)technology has given new life to ocean acoustics and underwater acoustic engineering,motivating a variety of new direction finding technologies based on acoustic vector signal processing.Development processes of the AVS technology as well as the direction estimation technology based on the AVS have been reviewed.The research status has been summarized.On this basis,the passive direction finding technology is studied in this thesis.Firstly,the physical foundation for direction estimation in acoustic vector field is introduced,including the general knowledge of underwater acoustic vector field and the measurement model of a single AVS.In particular,problem of how to measure the acoustic pressure at the geometric center of a combined vector sensor is studied.The rationality of the measurement method which is commonly used is proved mathematically.In this thesis,passive direction finding methods using a single AVS are divided into two categories,which are called the acoustic energy flux based approaches and the equivalent array based approaches respectively.For the former,a brief introduction is presented,and two kinds of complex acoustic intensity methods based on histogram are verified using the measured data.The latter can be further divided into two categories,which are called the parametric methods and the spatial spectrum estimation methods.Four kinds of classical spatial spectrum estimation algorithms and two kinds of classical parametric algorithms are implemented on a single AVS.Simulation experiments show that compared with the traditional acoustic pressure sensor array,a single AVS has the ability of non fuzzy direction finding in the whole space.A new parametric method,the joint multi-parameter estimation algorithm based on tensor decomposition,is proposed.By using the output data of a single AVS and its higher-order cumulants,a data tensor which accords with the parallel factor model is constructed.After decomposing the tensor,estimators of the two-dimensional directions of arrival and the frequencies of the sound sources can be acquired.This algorithm outperforms the estimating signal parameter via rotational invariance techniques(ESPRIT).A new pseudo spatial spectrum estimation method,the improved Capon algorithm based on the AVS array,is proposed.This algorithm takes advantage of the signal cancellation phenomenon of the minimum variance distortionless response(MVDR)beamformer to construct a pseudo spatial spectrum which is sensitive to whether the observed direction is equal to the expected one.The pseudo spectrum of the improved algorithm has more sharp peaks than the traditional spatial spectrum estimation methods,and performs better under the conditions of low SNR and snapshot deficiency.Finally,the normal mode theory,which is more suitable for the actual shallow water environment is studied,and the measurement model of the AVS array based on this theory is derived.Taking the multiple signal classification(MUSIC)algorithm as example,the influence of model mismatch on direction estimation is studied,and the normal model matched algorithm is presented.The original plane wave hypothesis makes the direction estimation result biased,while the normal model matched algorithm can realize unbiased estimation.