Abstract:Based on the three indicators of altitude angle, signal-to-noise ratio and pseudorange residual, this paper adopts K-means(Kmeans++), iterative self-organizing data analysis method(ISODATA) and density-based spatial classification with noise(DBSCAN) to classify the GNSS data in complex urban observation environments. We evaluate the classification accuracy of different algorithms using pseudorange single point positioning(SPP). The results show that the Kmeans++ algorithm has the best classification accuracy. The accuracy of positioning in three directions of E,N and U is 2.56 m, 3.25 m, and 9.73 m respectively; compared with not using the Kmeans++ algorithm, the positioning accuracy is improved by 57.86%, 47.64%, and 60.98%. To further verify the performance of the algorithm, the accuracy of the Kmeans++ algorithm is compared with the signal-to-noise ratio and height angle threshold algorithm. The results show that the plane and three-dimensional positioning accuracy of the Kmeans++ algorithm is significantly improved by 24.87%, 39.07%(signal-to-noise ratio algorithm) and 41.36%, 59.91%(height angle threshold algorithm), respectively.
LI Feixiang,ZHAO Lewen,TANG Geshi. GNSS Signal Classification and Accuracy Evaluation in Complex Observation Environment[J]. jgg, 2022, 42(8): 852-856.