The Study On Passive Tracking Of Underwater Moving Target

It is well known that the passive tracking methods of underwater moving target are various. The former researches are almost based on the state estimation model and estimate the target's motion elements indirectly by using the bearings-only, which is called as target motion analysis (TMA). The practical application of traditional bearings-only TMA is inconvenient because it needs a moving platform. Accordingly, many other TMA methods without maneuver are proposed using the Doppler frequency and all that. With the development of theory and technology, the matched field passive location and the bearings-only TMA based on association of multiple sensors have attached importance, including the research on the underwater GPS. The new methods contain many challenging theoretical problems.The paper focuses on the passive tracking of the vector buoy array and the involved theories are discussed deeply and thoroughly. In the aspect of general analysis, the study includes the calculation model, the postposition flatness of bearing-time sequence and the simulation evaluation of tracking. In order to prepare statistical technology for the property analysis of bearing estimation, the statistical property of single vector sensor must be studied first. The output model of vector sensor, the stability and ergodicity of the model, the first and second moments of vector sensor cross-spectrum estimation and its correlation moments are given. At last, the first-order probability density is derived. On the basis of the above, the bearing estimation property of vector sensor is discussed, including the probability density function and mathematical characteristic of bearing estimation, the influence of electrical rotation, bearing correction in the low SNR, the estimation method of SNR, the restriction of multiple targets bearing estimation, the postposition process and so on. The analysis of lake experiment's data validatethe supposing model of vector sensor and some theories, including the dropping noise-field analysis, the distance testing, the biasing phenomenon of bearing-time course and the spectrum comparison between acoustic intensity and pressure. The five kinds of estimation methods for spectrum level SNR are projected and suitable for the estimation of experimental environment.To pick up the same single-frequency signal from multi-channel, we can make use of multiple spatial sensors or collect the time-sharing data from the same sensor. Once the background noise is independent and the integral time is long enough, the cross-spectrum algorithm can depress the noise adequately and acquire the pure signal power spectrum theoretically. In terms of line spectrum, the cross-spectrum is sensitive to the phase strongly and the read signal power is weakened by the cosine factor of phase. Owing to phase difference, the target's maneuverability, the change of relative position and the uncertainty of time-sharing can result in the fluctuation of line spectrum. It's the probable reason why the cross spectrum is lack of the engineering foundation. Because the statistical property research of cross-spectrum about single vector sensor can't- be found in the literature about vector sensor and signal processing in the past ten years, the paper can't cite the similar conclusions.From the view of signal processing, the physical foundation of single vector sensor is that it is composed of four sensors in the same point, the phase differences between channels are stable, the provided quadrature process ensures the independence of noise and the projective relation contains the bearing information. Obviously, the discussion about cross-spectrum estimation property has general meanings and the bearing estimation of single vector sensor is special. The new theories can analysis experimental data skillfully and distill the various information of isotropic noise and signal.The single vector sensor is hard to resolve multiple targets of same-frequency, but can give the synthetic bearing at least. For the left-right ambiguity of line array, the...