大尺度磁性目标跟踪方法研究

Research on Large-scale Magnetic Target Tracking Method

针对磁定位中由于目标尺度不可忽略导致的观测模型不匹配问题,提出一种仅需2个三轴磁传感器测量数据,且能够自适应目标尺度的滤波估值型实时磁定位算法。该方法基于最大似然选择思想,建立了不同个数磁偶极子构成的均匀线列阵模型作为候选观测模型;推导了一种适用于高维非线性动态参数估计的随机高斯流滤波算法,克服了大尺度目标模型高维度和强非线性导致的性能下降问题,提高了实时定位精度。同时,采用该算法对各个偶极子线阵模型分别进行滤波解算,选取最大似然值对应的滤波结果作为当前时刻估计值,实现了对目标尺度的自适应。通过双层壳体潜艇有限元磁场仿真模型数据对算法的有效性进行了验证,结果表明,相比现有方法,该算法估值更为准确,能够满足精确实时定位目标的需求,且具有更好的收敛性能。

To address the mismatch problem of observation model caused by non-ignorable target scale in magnetic positioning,a real-time magnetic positioning algorithm of filter estimation adaptive to target scale only based on the measurement data by two three-axis magnetic sensors is proposed.Based on maximum likelihood selection idea,the uniform line array model composed of different numbers of magnetic dipoles are established as the candidate observation model,then a stochastic Gaussian filtering algorithm suitable for the estimation of high dimensional nonlinear dynamic parameters is deduced.The performance degradation problem caused by high dimension of large-scale target model and strong non-linearity is overcome,the real-time positioning accuracy is improved.Meanwhile,the algorithm is utilized to filter and solve each dipole linear array model respectively.The filtering results corresponding to the maximum likelihood value are selected as the current time estimate,the self-adaptation of target scale is realized.Finally,the validity of the algorithm is verified by finite element magnetic field simulation data of double shell submarine,the results show that the proposed algorithm is more accurate and can meet the accurate and real-time requirement for target positioning with better convergence performance compared with the current method.