基于循环神经网络算法的 室分用户定位与感知监控方法

本文提出了一种基于机器学习的室分用户位置定位方法，利用循环神经网络算法构建楼宇立体栅格的关联模型，再结合室分用户的MR数据，完成室分用户位置轨迹的拟合与位置定位。在此基础上，关联信令侧KQI感知数据实现室分用户感知的智能监控。在某市进行试点，对识别出的用户感知问题现场排查确认，位置定位精确度达到88.65%。经过两个月的专项整治，解决用户感知问题327处，室分MR覆盖率由94.41%提升至97.22%。

A method of user location and perception monitoring based on cyclic neural network algorithm

In this paper, a machine learning-based location location method for compartment users is proposed. The circular neural network algorithm is used to construct the building three-dimensional raster association model. Then, combined with the MR data of compartment users, the location trajectory fitting and location location of compartment users are completed. On this basis, KQI sensing data on the correlation signaling side can realize intelligent monitoring of user perception. A pilot project was carried out in a certain city to investigate and confirm the identified user perception problems on the spot, and the location positioning accuracy reached 88.65%. After two months of special rectification, 327 user perception problems were solved, and the room coverage rate of MR increased from 94.41% to 97.22%.