Research Of Data Fusion Based On Vector Hydrophones

Multi sensor data fusion is a technology of multi subject across. Recent twenty years, technology of multi sensor data fusion is generally attended and widely used along with technologies of sensor,computer,communication and signal processing developing.Vector sensor is combined with pressure sensor and velocity sensor, and it has characteristics of multi sensor. The paper starts with vector hydrophone, circumfuses signal processing methods of vector sensor, applies technology of data fusion into it, exerts the best of its detecting ability, so as to vector sensor can apply to wider fields.The main research content can be summarized as follows:(1) The relations between pressure and velocity measured by vector sensor are in detail analyzed. The basic bearing methods of single vector sensor are introduced, the models of two orders and high-level statistical quantum of vector sensor are established, many bearing methods are discussed, such as: arc sine function method,arc cosine function method and pressure associating with velocity method, and their performances are analyzed. The time-domain fusion model and the frequency-domain fusion model are established. The frequency-domain fusion estimating methods are detailedly discussed on the rule of maximum likelihood. The frequency precision is improved by technology of ZFFT for line spectrum signals, and effective frequency bound is searched by using ant algorithm for broadband signals. The results of computer simulation and analysis of data from outfield trials show that fusion algorithm excels traditional direct estimation and can improve estimation capability of vector sensor.(2) Multi targets distinguishing method based on reciprocal spectrum acoustic intensity is simply introduced. A new method based on statistical quantum is raised, and it can be solved by improved genetic algorithm. Among different signal models(single frequency signah broadband gauss signah non-gauss signal), the solving forms of the algorithm are analyzed, the methods of interferential signals separated by single vector sensor are discussed. The results of computer simulation indicate that the algorithm can effectively distinguish and track the bearing of multi targets which aren't interferential, and that it can distinguish lesser targets on frequency-domain when the interferential coefficient is higher, and that it can separate frequency spectrum of interferential signals.(3) Nonlinear multi sensor fusion algorithms are simply discussed. Optimal weighed fusion algorithms for vector sensor bearing are researched, every weight computing methods are detailedly summarized. The fusion detected method is raised, and it can improve the precision of bearing based on vector sensor and can eliminate the effect of detecting precision on account of target azimuth.(4) The relation matrixes of vector sensor are established according to the synthetically supported degree of pressure hydrophone,velocity hydrophone and every vector sensors. The relation matrixes indicate a certain extent relations of each sensor. On the foundation of it, the vector fusion algorithm based on relation matrix is raised, and it improves multi vector sensor fusion results.(5) The technology of vector sensor state spatial fusion is researched. For the cooperative target whose velocity is known, the methods of estimating the distance of target are discussed (acoustic pressure,sound intensity,single direction sound intensity and azimuth estimated), and the ranging errors of the four methods are analyzed on SNR and azimuth. On the foundation of it, fusion algorithm is raised to estimate distance, and the precision is improved. The model of target moving is established, the method of combining acoustic intensity and azimuth is studied, and its errors are analyzed. The bearing only positioning precision based on single platform is lower, and the state of target can't be observed in some instances. There are two methods to solve the problem: the first one is to combine vector sensor and ranging equipment; the second one is to fusion position by multi platforms. Spatial fusion for multi platforms can be transformed to position of two platforms. The paper is on the foundation of two platforms positioning, the technology of multi sensor fusion bearing and positioning is based on least-square criterion. The simulation results based on extended kalman filter indicate that the bearing only method based on two platforms has preferable observability, and it can accurately truck the target trail.The technology of vector sensor data fusion is researched in the paper. The fusion theory is attested by the results of computer simulation and some trials.