采煤机精准定位一直是多年来亟需解决的瓶颈，是实现井下采煤机智能化、无人化开采的关键。针对当前采煤机定位精度较低的问题，提出将超宽带 (Ultra-Wideband ,UWB)基站安装在综采工作面端头，当采煤机运行至端头时，由UWB定位系统为惯性导航端头校准提供基准。为使UWB定位系统获得较高的定位精度，考虑到基站位置误差、尺度因子误差和偏差对定位精度的影响。提出了一种校准方法，能够独立计算这些参数，对基站的位置误差进行校准补偿；由于非视距(Non-Line-of-Sight,NLOS)环境的影响，NLOS偏差因距离变化而有所不同，同时考虑到测量噪声的不确定性和时变性，提出了变分贝叶斯无迹卡尔曼滤波(Variational Bayesian unscented Kalman filter ,VBUKF)对NLOS测距信息进行平滑处理，减小NLOS误差的影响，提高距离估计精度。在校准补偿和VBUKF平滑基础上，采用Caffery定位方法(Caffery localization,CL)解算目标节点的坐标，为进一步提升定位精度，采用高斯-牛顿迭代算法(Gaussian-Newton,GN)对CL算法的结果进行优化，并通过UWB定位系统进行动静态试验验证所提方法的有效性。试验结果表明:VBUKF平滑能够有效地减小NLOS误差，GN算法能够进一步提升定位精度；经过校准补偿和VBUKF平滑后，X轴，Y轴及Z轴的平均定位误差大大减小，CL算法和CL-GN算法的平均定误差分别由0.347、0.250 m减小为0.239、0.109 m,平均定位精度分别提高了31.1%,56.4%；同时动态运动轨迹更加接近实际轨迹，表明该方法能提升UWB系统的定位精度，能够为井下定位技术提供理论参考，但是该方法有待进一步现场试验的验证。

The precise positioning of the shearer has been a bottleneck that needs to be addressed urgently for many years, and it is the key to realize the intelligent and unmanned mining of the underground shearer. In view of the issue of the low positioning accuracy of the shearer, it is proposed to install the Ultra-Wideband (UWB) base station at the end of the fully-mechanized mining face. As the shearer runs to the end, the UWB positioning system provides the reference for the calibration of the inertial navigation. In order to obtain higher localization accuracy for UWB positioning system, the influence of base station position error, scale factor error and deviation on positioning accuracy is considered, and a calibration strategy is proposed which can independently calculate these parameters to calibrate and compensate the position error of the base station. The non-line-of-sight (NLOS) bias varies due to the impact of NLOS environment, with the consideration of the uncertainty and time-variant measurement noise, and the variational Bayesian unscented Kalman filter (VBUKF) is proposed to smooth the NLOS measurement distance to reduce the effect of the NLOS error and ameliorate distance estimation accuracy. On the basis of calibration compensation and VBCKF smoothing, the Caffery localization (CL) approach is utilized to determine the coordinates of the target node. To further improve the positioning accuracy, Gaussian-Newton localization algorithm (GN) is employed to optimize the results of the CL result, and the effectiveness of the proposed method is verified by dynamic and static experiments using UWB positioning system.The experimental results show that the VBUKFsmoothing can efficiently reduce the NLOS error, and the GN algorithm is able to further improve the positioning accuracy. The positioning errors on the X-axis, Y-axis, and Z-axis are significantly diminished after calibration and VBUKF smoothing. The average fixed error of CL and CL-GN method is decreased from 0.347 m, 0.250 m to 0.239 m, 0.109 m, respectively, and the corresponding average positioning accuracy is improved by 31.1% and 56.4% respectively. Simultaneously, the dynamic motion trajectory is closer to the real trajectory, which certifies that the proposed approach is capable of improving the positioning of UWB system and providing theoretical guidance for underground positioning technology, but this scheme requires to be certified by further actual environmental verification.