Mine weighted centroid positioning algorithm based on improved Gaussian mixture filter

In view of the problems that RSSI positioning algorithm has large ranging error and inaccurate positioning result due to non-line-of-sight propagation of wireless signals in complex environment of underground coal mine, a mine weighted centroid positioning algorithm based on improved Gaussian mixture filter was proposed. Firstly, corresponding RSSI measurement data of unknown nodes are clustered according to the maximum expectation algorithm, and the data are divided into multiple Gaussian probability density functions models. Then, according to the characteristics of the data, the sampling data is optimized by using Akaike information criterion, and the accurate measured values are obtained. Finally, the initial coordinates of the unknown node are calculated, the error between the initial coordinates of the unknown node and the real coordinates are used as the weight factor, and the final coordinates of the unknown node are calculated combining with the centroid positioning algorithm, so as to realize target positioning. The simulation and experimental results show that the positioning algorithm can realize high precision personnel positioning in underground coal mine, and the average positioning error is 1.83 m.