Study Of Track-before-detect Method Based On The Particle Filter For Underwater Target Detection

Because electromagnetic waves and light transmission attenuation in the water transmission attenuation is large, and sound can travel very long distances in the ocean, underwater acoustic wave is a major way of the information transmission, underwater target detection, underwater imaging, etc. With the development of acoustic stealth technology, and the complexity of the underwater acoustic channel, underwater target detection is facing the problem of "two one-thousand". One is that the number of interferences is one thousand times largers than the targets. The other is that the strength of interferences is more than one thousand times stronger than the targets. These bring great challenge to underwater weak target detection. Track before detect (TBD) method is an important research direction of the weak target detection, it is different from the traditional track after detect (TAD) method. This method is not to set a threshold to determine the existence of the target, but by relating the physical movement characteristics of target, and mining useful information from the original data, to extract target information through the data accumulation. Particle filter (PF) is an important means for dealing with nonlinear sequential filtering, and its sequential feature is suitable for target tracking. This article presents the TBD method based on the PF for underwater weak target detection. The main work is as follows.First, sequential Bayesian filters are employed to track underwater targets. A state space model is established for underwater target detection, which includes a target motion model and an observation model with a vertical linear array. In terms of the nonlinear characteristics of underwater sound propagation, the extended Kalman filter (EKF), the ensemble Kalman filter (EnKF) and the PF are discussed. The simulation analysis shows the feasibility of using PF to track underwater targets.Secondly, the matched field processing (MFP) is employed to establish the TBD processing model of acoustic passive target detection. The principle of the PF for tracking before the detection is studied. Then, the PF-TBD algorithm is put forward for underwater weak target detection and tracking. The simulation analysis shows the algorithm is feasible, and it is superior to the traditional PF.Finally, a sea test was carried out. In the test we took the autonomous underwater vehicle (AUV) as the passive target for tracking. A vertical array was used to receive the signal radiated by the AUV. The experimental results show that the PF-TBD algorithm can effectively track the AUV.