基于序贯集中式水下扩展阵浮标动态非线性系统融合技术研究

项目名称： 基于序贯集中式水下扩展阵浮标动态非线性系统融合技术研究

项目编号： No.51309191

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 水利工程

项目作者： 苟艳妮

作者单位： 西北工业大学

项目金额： 25万元

中文摘要： 对水下目标进行有效的定位跟踪是浮标探测的主要任务，由于受到海洋洋流等因素干扰，浮标网络在洋流中的位置随机性偏差较大，同时，海水的各向异性和声波在水中传播的非线性，使得实际的浮标动态系统定位跟踪精度很差。传统的卡尔曼滤波算法对于海洋环境中的强非线性系统，估计误差会进一步扩大甚至发散，已不能满足高精度水下目标跟踪性能要求。本项目引入扩展阵声纳浮标作为系统基本单元，确保在一定的高测向精度条件下，深入探讨水下动态非线性系统目标估计融合问题。在针对浮标网络在海洋环境位置不确定性和水下量测受限等影响目标定位跟踪精度的实际动态问题中，以集中式多传感器目标估计融合理论为主要技术手段，建立非线性系统下的无迹卡尔曼滤波框架，并结合序贯融合思想，将序贯集中式融合算法引入到无迹卡尔曼滤波算法当中，旨在保证一定计算复杂度的前提下，获得较为精确的估计结果，并期望通过水池和湖上试验进一步验证新算法的正确性和有效性。

中文关键词： 扩展阵浮标；模拟退火定位；非线性滤波；集中式融合；多站址信息处理

英文摘要： Effective positioning and tracking underwater target is the main task of detection buoys. Due to interference by the ocean currents and other factors, the randomness position deviation of the buoy network in the ocean currents is larger, meanwhile, because of the seawater's anisotropy and the acoustic non-linear propagationn in the water, the detection accuracy of the actual buoys dynamical system is poor. For strongly nonlinear systems in the ocean environment, it is estimated that the error of traditional Kalman filter algorithm will further expand even divergence, so the algorithm can no longer meet the purpose of accurate detection of modern war. This project introduced extended sonar buoys as a basic unit of the network to ensure certain high-finding accuracy to explore the problem of target detection. For the actual detection problems which is caused by the uncertainty of the buoy network in the marine environment and the underwater measurement limitation, the centralized multi-sensor target estimation fusion theory is adopted as the main technical means, the establishment of a nonlinear system under the unscented Kalman filtering framework, combined with sequential fusion theory, and the sequential centralized fusion algorithm is also introduced to the adaptive Kalman filter algorithm which aims to obtain

英文关键词： extended buoy；simulated annealing localization；nonlinear filter；centralized fusion；multi-senser information processing