Abstract: In order to obtain the accurate positions of firefighters in a complex fire environment, an ultra-wideband（UWB） based indoor collaborative positioning algorithm for firefighters is proposed, which makes full use of the ranging information from target to UWB base stations and other targets for positioning. Firstly, a linear fitting method is used to preprocess the standard deviations in the measurement distances. Then, aiming at the problem of target position calculation and non-line-of-sight（NLOS） error mitigation, a collaborative positioning algorithm based on biased extended Kalman filter is proposed. According to the internal connection between the targets to be located, the new state equation and measurement equation are established, and the Kalman gain is adjusted by the constructed coefficient matrix to correct the deviated position estimation values. Finally, for the problem of positioning coordinates jump, an improved mean filter algorithm based on threshold screening is proposed to optimize the positioning results secondarily. The experimental results show that the proposed algorithm has a positioning accuracy of up to 0.17 m in weak NLOS environment and 0.28 m in strong NLOS environment. Compared with other algorithms in this paper, the proposed algorithm has better positioning performance, reduces the requirement of high distribution density of UWB base stations, uses the resources of the whole collaboration network to the greatest extent, and provides a solution to the problem of difficult or inaccurate positioning caused by obstacles for the firefighters group in the fire scene.

Key words: ultra-wideband（UWB）, firefighters positioning, collaborative positioning, extended Kalman filter, mean filter

摘要： 为了在复杂火场环境下获取消防员的精确位置，提出基于超宽带（ultra-wideband，UWB）的消防员室内协同定位算法，充分利用目标到UWB基站以及到其他目标的测距信息进行定位。采用线性拟合方式对测量距离中存在的标准偏差进行预处理；针对目标位置解算及非视距（non-line-of-sight，NLOS）误差缓解问题，提出基于偏移扩展卡尔曼滤波的协同定位算法，根据待定位目标之间的内在联系，建立新的状态方程和量测方程，并通过构造的系数矩阵调整卡尔曼增益，修正偏离的位置估计值；针对定位坐标跳变问题，提出基于阈值筛选的均值滤波算法对定位结果进行二次优化。实验结果表明，所提算法的定位精度在弱NLOS环境下高达0.17?m，在强NLOS环境下高达0.28?m，与文中其他算法相比具有更好的定位性能，降低了定位对UWB基站分布密度高的要求，最大程度地使用了整个协同网络的资源，为深入火场内部的消防员群体因障碍物遮挡导致的定位困难或定位不准问题提供了一种解决方案。

关键词: 超宽带（UWB）, 消防员定位, 协同定位, 扩展卡尔曼滤波, 均值滤波