一种随机森林与深度学习结合的室内定位方法

为更加实时、精确地识别运输设备的位置信息和特殊货物的位置信息尤其是朝向信息以提高工作效率，利用仿真实验对室内物体进行定位和朝向判断的探究。利用天线阵列布置室内环境，在考虑电磁波极化特性的基础上利用信道传播模型进行建库；利用随机森林进行朝向判断后通过不同的深度学习模型进一步实现定位。实验结果表明：该模型不仅能实现朝向判断，而且其定位误差比仅利用深度学习模型的定位误差降低约0.14 m。

An Indoor Positioning Method Based on Random Forest and Deep Learning

In order to identify the position information of transportation equipment and the position information of special goods, especially the orientation information, so as to improve the working efficiency, the simulation experiment is used to explore the position and orientation of indoor objects. The indoor environment is arranged by the antenna array, and the channel propagation model is used to build the database on the basis of considering the polarization characteristics of electromagnetic waves. The random forest is used to judge the orientation and then the location is further realized by different deep learning models. Experimental results show that the model can not only achieve orientation judgment, but also its location error reduces by about 0.14 m compared with that only using the deep learning model.