基于DTW的地磁指纹定位方法

Geomagnetic Fingerprint Localization Method Based on DTW

针对动态时间规整匹配(DTW)的奇异性问题,利用克里金插值方法(Kriging)基于156个参考点插值得到545个指纹点建立后台指纹库以减少离线采集的工作量,采用二次加权质心算法对DTW进行改进,首先利用一次加权质心算法获得结果位置点的初始位置,再用二次质心加权算法对定位结果位置点进行偏远误差点的剔除,最终获得可靠性较高的定位结果。实验表明,利用Kriging可以节省71.4%的工作量;在走廊和大厅的环境中定位精度分别可以达到2.01m和4.19m。改进的DTW在长廊和大厅环境中的平均定位误差分别为1.64m和2.74m,较原算法在定位精度上有明显提升。

Aiming at the singularity problem of dynamic time warping matching(DTW),this paper uses Kriging interpolation method to obtain 545 fingerprint points based on 156 reference points interpolation to establish a background fingerprint database to reduce the workload of offline collection.The centroid algorithm improves the DTW.Firstly,the initial position of the result position point is obtained by a weighted centroid algorithm.Then,the second centroid weighting algorithm is used to remove the remote error point of the positioning result point,and finally obtain the highly reliable positioning result.Experiments show that using Kriging can save 71.4%of the workload;in the corridor and hall environment,the positioning accuracy can reach 2.01 m and 4.19 m respectively.The average positioning error of the improved DTW in the corridor and hall environment is 1.64 m and 2.74 m,respectively,which is significantly improved compared with the original algorithm.