煤矿井下基于Attention机制的LSTM地磁定位算法

针对利用无线射频信号对井下人员定位受环境影响大的问题，本文提出了基于Attention机制的LSTM地磁定位算法。利用矿用智能手机以路径连续采集方式获取井下地磁数据，通过分析井下巷道地磁信号变化的空间差异性和时间稳定性，建立面向井下拓扑结构的地磁指纹数据库；在LSTM模型中引入Attention机制，使其能够根据地磁序列与位置的相关性进行训练且给予不同的权重；将井下实时采集的地磁数据输入模型后进行位置估计。试验结果表明，通过神经网络LSTM学习地磁序列与相应位置的映射关系，可有效提高定位精度。

LSTM geomagnetic positioning algorithm based on Attention mechanism in underground coal mine

Aiming at the problem that the use of radio frequency signals for underground personnel positioning is greatly affected by the environment, this paper proposes an LSTM geomagnetic positioning algorithm based on the Attention mechanism. The mine smart phone is used to acquire underground geomagnetic data in the way of continuous acquisition of paths. By analyzing the characteristics of spatial difference and time stability of the geomagnetic signal changes in underground roadways, a geomagnetic fingerprint database oriented to the underground topology is established. The Attention mechanism is introduced into the LSTM model to enable which can be trained and given different weights according to the correlation between the geomagnetic sequence and the position; the geomagnetic data collected in real time downhole is input into the model for position estimation. The experimental results show that learning the mapping relationship between the geomagnetic sequence and the corresponding position through the neural network LSTM can effectively improve the positioning accuracy.