煤矿井下基于降噪自编码器的航向估计算法研究  

Research on Heading Estimation Algorithm Based on Denoising Auto-Encoder in Coal Mine

作　　者：郭倩倩 崔丽珍[1] 杨勇[1] 高丽丽[1] 赫佳星[1] GUO Qianqian;CUI Lizhen;YANG Yong;GAO Lili;HE Jiaxing(School of Information Engineering,Inner Mongolia University of Science and Technology,Baotou Inner Mongolia 014010,China)

机构地区：[1]内蒙古科技大学信息工程学院,内蒙古包头014010

出　　处：《传感技术学报》2021年第6期797-803,共7页Chinese Journal of Sensors and Actuators

基　　金：国家自然科学基金项目(61761038);内蒙古自然科学基金项目(2020MS06027);内蒙古自治区科技计划项目(2019GG328)。

摘　　要：针对煤矿井下高动态、强磁场的复杂环境,使用智能手机中陀螺仪解算航向角存在较大累积误差这一问题,提出一种基于降噪自编码器的卡尔曼滤波融合航向估计算法.该算法由卡尔曼滤波融合航向解算和旨在消除原始惯性传感器数据噪声的降噪自动编码器组成,通过训练降噪自编码器对井下惯性传感器数据进行降噪处理,并利用卡尔曼滤波融合陀螺仪积分航向解算和九轴传感器航向解算,得到矿工运动的航向角.本文在鄂尔多斯高头窑煤矿采集矿工运动数据,试验结果表明,在高动态复杂矿井下,本文算法较九轴传感器航向解算有较强的抗干扰能力,满足井下矿工PDR航向估计需求.In view of the complex environment of high dynamic and strong magnetic field in coal mine,there is a large cumulative error in using gyroscope to calculate heading angle in smart phone.In this paper,a Kalman filter fusion heading estimation algorithm based on denoising autoencoder is proposed.The algorithm is composed of Kalman filter fusion heading solution and denoising autoencoder which aims to eliminate the original inertial sensor data noise.Through training denoising autoencoder,the underground inertial sensor data is denoised.Kalman filter is used to fuse gyro integral heading calculation and nine axis sensor heading calculation to obtain the heading angle of miners’movement.In this paper,the movement data of miners are collected in Gaotouyao coal mine of Ordos.The experimental results show that the algorithm in this paper has stronger antiinterference ability than nine axis sensor heading calculation in high dynamic complex mine,which can meet the demand of PDR course estimation of underground miners.

关 键 词：煤矿井下 降噪自编码器 航向解算 四元数 卡尔曼滤波 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置]