Underground personnel positioning algorithm based on clustering and K-nearest neighbor algorithm

In view of problems of large amount of calculation, low real-time performance and low positioning accuracy of existing fingerprint-based underground positioning algorithm, underground personnel positioning algorithm based on clustering and K-nearest neighbor algorithm was proposed. Bisecting k-means clustering algorithm is used to classify collected RSSI data to establish an offline fingerprint database. Real time RSSI values are collected by wireless mobile terminal and dynamic corrector and stored in online positioning database and dynamic correction database respectively. According to offline data and real-time data, weight value is calculated using software and hardware dynamic correction weighted K-nearest neighbor algorithm, and real-time position is estimated by combining the physical location information of the point to be measured in the offline fingerprint database. The example analysis results show that the minimum standard error of the proposed positioning algorithm is 0.46 m, the maximum standard error is 3.26 m, and the average error is 1.62 m. The results of comparative analysis show that the proposed algorithm has higher precision and better real-time performance than the algorithm without clustering analysis. Compared with the algorithm without dynamic correction of weights, the computation time of the proposed algorithm is slightly increased, but the positioning accuracy is increased by 37.21%.