复杂环境下BDS场景精度等级聚类模型

Clustering model of BDS scene accuracy level in complex environment

针对实际复杂场景下北斗卫星导航系统(BDS)导航定位精度不足问题,提出一种基于美国国家海洋电子协会(NMEA)协议的BDS自适应场景迭代自组织数据分析技术(ISODATA)聚类算法,以辅助BDS终端针对不同场景采取合适精度的补偿策略,即从实时NMEA报文提取与BDS定位相关的环境特征指标进行场景聚类分析,依据BDS精度等级对导航场景进行自动分类和精度匹配,并采用有监督的k近邻算法(kNN)对ISODATA场景聚类结果的准确度进行评估。实验结合(上海)复杂城市环境下的实际路测数据,提取BDS导航场景聚类模型结果及相应的匹配精度等级,给出谷歌(Google)地图上再现7类典型场景下不同路段对应的12种自动聚类匹配精度,以及相应的定位精度偏差均值。实验结果表明:此聚类方法的均方误差多次迭代后效果明显优于其他聚类方法,根据本次路测可识别的精度聚类等级范围(最近0.02 m、最远约3 m)给出精度补偿的相对应的建议,也验证了模型的有效性。

Aiming at the problem of insufficient positioning accuracy of Beidou navigation in real complex scenes,a BDS adaptive scene ISODATA clustering algorithm based on NMEA protocol is proposed to assist BDS terminal to adopt appropriate compensation strategy for different scenarios.That is,the scene feature clustering is extracted from the real-time NMEA message and the environmental feature indicators related to BDS positioning,and the navigation scene is automatically classified and matched according to the BDS accuracy level,and the supervised KNN method is used to accurately cluster the ISODATA scene.The experiment combines the actual road test data in complex urban environment(Shanghai),extracts the results of the BDS navigation scene clustering model and the corresponding matching accuracy level,and gives 12 kinds of automatic clustering corresponding to different road segments in 7 typical scenes reproduced on google map.Matching accuracy,and the corresponding positioning accuracy deviation mean.The experimental results show that the mean square error of this clustering method is obviously better than other clustering methods after multiple iterations.According to the accuracy of the road test,the clustering level range(the nearest 0.02 meters,the farthest is about 3 meters)The corresponding recommendations for accuracy compensation also verify the validity of the model.