A Study On Low Target Localization Filter Algorithm And Data Association Algorithmon Acoustic Detection

The passive acoustic detection technology is a passive detection technology.Compared with the active radar, passive acoustic sensors have lower energyconsumption, small volume, small quality and higher concealing ability, etc. As apowerful complement to active localization system, it has an important applicationvalue and application prospect. Under the pre-research projects of“11thFive-YearPlan”, the paper researches the filter algorithm and data association algorithm of thetarget location under the passive acoustic detection network. The major research workand innovation are as follows:1. The background and present status of the research in the filed of the passivedetection technology and the passive acoustic detection network arepresented. Some typical target tracking algorithms and data associationalgorithms are reviewed, including the Kalman Filter algorithm, theExtended Kalman Filter algorithm, the Particle Filter algorithm, the NearestNeighbor algorithm, the Probabilistic Data Association algorithm, and theJoint Probabilistic Data Association algorithm, etc.2. According to the problem of single target tracking under passive acousticdetection network, a passive acoustic sensor network time-delay targetlocalization algorithm based on MVEKF is presented. Firstly, in order tosolve the problem of single acoustic sensor can’t completely observe thetarget states, a method of use multi-acoustic sensors to form a network isgiven. Secondly, according to the characteristic of existing time-delay inmeasurement information, an angle correction method is presented. At last,some common localization coordinates are introduced. A rectangular planecoordinate system is built for the passive acoustic detection network. Thevalidity of this algorithm is verified through some simulation experiments oftarget motion scene.3. According to the problem of large quantities of calculation in traditionaljoint probability data association algorithm, and hard to meet thecharacteristics of real-time, a cheap joint probability data association ispresented. Under the background of multi-targets track and location in passive acoustic detection network, combined with the unscented KalmanFilter algorithm, a unscented cheap joint probability data association ispresented. The validity of this algorithm is verified through some simulationexamples of target motion scene.4. According to the problem of the validity of the proposed localizationalgorithm in the true environment, using the actual measured data providedby the cooperative unit, the proposed algorithms are tested and verified onthe passive acoustic detection network data fusion platform. Through thedetailed performance comparison and analysis, the proposed algorithm iscompletely meet the requirements of the indicators in the actual project.