基于多尺度贯序式卡尔曼滤波的运动声阵列跟踪算法

Tracking Algorithm of Motion Acoustic Array Based on Multi-scale Sequential Kalman Filter

为了研究运动声阵列对二维目标在复杂环境中的实时跟踪性问题,根据运动声阵列及二维目标的运动特点建立了状态方程与测量方程,并将其描述为块的形式.根据不同的状态块,利用小波变换把状态块分解到不同尺度上,分别在时域和频域上建立相应尺度上的状态与观测信息之间的关系;采取卡尔曼滤波器递推思想来实现运动声阵列的多尺度贯序式卡尔曼滤波算法,根据最小二乘误差估计理论推导了运动声阵列跟踪系统在球坐标系和直角坐标系下的误差,为提高系统跟踪精度奠定了理论基础,并为工程应用提供了实际方法.与传统的卡尔曼滤波算法相比,Matlab仿真结果表明了本文算法的有效性和优越性.

In order to study the motion acoustic array＇s real-time tracking to two-dimensional target in complex environment, the state equation and measurement equation based on the motion characteristics of dynamic acoustic array and two-dimensional target are established and converted into block form.Then,the state blocks are assigned onto different scales by wavelet transform.The relationship between the state and the measurement information in corresponding scale is established in time domain and frequency domain.After that,the multi-scale sequential Kalman filter algorithm is obtained based on the Kalman filter recursive theory,and the errors of motion acoustic array tracking system in spherical coordinates and rectangular coordinates are deduced by least square error estimation,which lays the theoretical foundation for improving the system tracking precision and provides a practical method for application.Compared with the traditional Kalman filter algorithm,the presented algorithm shows its validity and superiority in Matlab simulation.