基于改进粒子滤波的井下跟踪算法研究与实现

井下环境复杂多变，射频信号易受到阴影效应、多径衰落等因素的影响。采用传统的粒子滤波跟踪方法误差较大，研究了一种基于改进粒子滤波的井下跟踪算法。初始化阶段利用第一次指纹匹配算法的定位结果来设计初始化概率分布函数；采用核函数法与指纹匹配技术相结合的算法，在采样数据中搜索与目标节点指纹特征相匹配的位置并加权得到位置坐标作为跟踪中的观测值；最后利用粒子滤波将观测值与目标运动状态相融合以跟踪目标运动轨迹。实验结果表明，粒子滤波算法较优化卡尔曼滤波算法更适用于井下跟踪；改进的算法有效增强了跟踪系统的可靠性，提高了跟踪精度，满足了井下的跟踪要求。

Research and Implementation on Underground Tracking AlgorithmBased on Improved Particle Filter

The downhole environment is complicated, RF signal is influenced by shadow effect, multipath fading, etc. Tracking error of using traditional method is bigger, An improved downhole tracking algorithm was studied, which optimized the particle filter algorithm. In initialization phase, first positioning results of fingerprint matching algorithm was used to design initial probability distribution function; by using the method that the kernel function method combined with fingerprint matching technique, searched the location which matched with fingerprint feature of the target node in the sampled data and put the weighted coordinates as observed values in the track; finally by utilizing particle filter to combine observed values with target motion state, tracked the target motion trajectory. The experimental results show that, particle filter algorithm is more suitable than improved kalman filtering algorithm for the coal mine tracking. The improved algorithm effectively enhances the reliability of the tracking system, increases the tracking accuracy and satisfies the requirement of underground track.