摘要
针对室内救援场景内救援人员精准定位困难的问题,设计了一种利用长短期记忆网络模型(Long Short-Term Memory,LSTM)与创新改进的融合启发式漂移消除算法(Improved Fusion Heuristic Drift Elimination,IFHDE)辅助惯性导航系统的室内定位方案。开展基于LSTM的零速检测算法设计,通过模型的学习能力,自适应调整阈值,提高零速检测精度,消除固定阈值带来的定位误差累积;创新改进融合启发式漂移消除算法,通过区分不同运动状态,设置修正条件,能够针对性地修正航向角误差,从而提高定位精度。通过在惯性导航系统框架中融合LSTM算法和IFHDE算法,执行量测更新实现对人员解算轨迹的精确还原。实验结果表明,与固定阈值的零速检测算法对比,LSTM零速检测算法在混合运动模式下的定位误差显著降低,降幅达到24.2%至46.9%;融合IFHDE算法后,定位误差进一步下降,降幅为9.2%至14.4%,误差占比不超过总路线的0.7%,能够满足救援人员室内精准定位的需求,提高轨迹还原精度。
To address the critical challenge of precise first responder localization in indoor emergency rescue scenarios,this study proposes,an indoor positioning scheme was designed by using the Long Short-Term Memory(LSTM)model and an innovatively improved fusion heuristic drift elimination algorithm(IFHDE)to assist the inertial navigation system.A zero-velocity detection algorithm based on LSTM was developed.Through the learning ability of the model,the threshold was adaptively adjusted to improve the accuracy of zero-velocity detection and eliminate the cumulative positioning error caused by the fixed threshold.The IFHDE algorithm was innovatively improved.By distinguishing different motion states and setting correction conditions,the heading angle error can be corrected specifically,thereby improving the positioning accuracy.By integrating the LSTM algorithm and the IFHDE algorithm in the inertial navigation system framework and performing measurement updates,the precise restoration of the personnel's calculated trajectory was achieved.Experimental results show that compared with the zero-velocity detection algorithm with a fixed threshold,the LSTM zero-velocity detection algorithm significantly reduced the positioning error in mixed motion modes,with a reduction of 24.2%to 46.9%.After integrating the IFHDE algorithm,the positioning error further decreases by 9.2%to 14.4%,and the error ratio does not exceed 0.7%of the total route,which can meet the demand for precise indoor positioning of rescue personnel and improve the accuracy of trajectory restoration.
作者
张明月
彭程
卜祥丽
黎冠
余星辰
汪洋
ZHANG Mingyue;PENG Cheng;BU Xiangli;LI Guan;YU Xingchen;WANG Yang(School of Electronic and Information Engineering,North China Institute of Science and Technology,Yanjiao,065201,China;School of Emergency Equipment,North China Institute of Science and Technology,Yanjiao,065201,China;Ministry of Emergency Management Big Data Center,Beijing 100010,China)
出处
《华北科技学院学报》
2025年第3期67-76,87,共11页
Journal of North China Institute of Science and Technology
基金
中央高校基本科研业务费资助项目(3142024036,3142024037,3142018048)
廊坊市科学技术研究与发展计划自筹经费项目(2024011066)。
关键词
室内定位
零速检测算法
航向修正算法
惯性导航
indoor positioning
zero-speed detection algorithm
course correction algorithm
inertial navigation
