煤矿井下人员定位系统对煤矿安全生产有着重要意义，井下巷道多呈狭长且环境复杂多变。单一定位方式无法满足多种条件下定位需求，且需部署新设备成本高，其中Wi-Fi随着矿用无线通信系统发展逐步实现井下全覆盖，而无需新设备。针对井下特殊环境，开展超宽带UWB测距与Wi-Fi指纹在井下人员定位上的应用研究，并提出一种以惯性传感器为辅的多元融合定位方法。主要内容如下：

（1）针对UWB设备时钟漂移和井下环境复杂问题，改进传统的TWR双程测距方法减少时钟漂移误差，并采用多基站协同通讯方式减少通信次数缩短定位时间，同时增设环境偏置系数，利用基站间测距情况对标签测距结果初步修正。然后针对人员运动过程中信号传输的多状态模式，建立高斯混合模型GMM分解测距信息，识别出视距条件下的测距结果并进行重构，得到最接近真实的距离信息。最后提出改进坐标解算算法，设计两阶段最大似然法估计位置坐标。实验结果表明，测距精度提升了68.5%，视距条件下静态平均定位误差0.28m，在定位精度和稳定性上均优于其他定位算法，为快速采集指纹标签提供基础。

（2）井下Wi-Fi指纹定位应用存在基站稀疏分布、特征度稀少等问题，采用信道状态信息CSI作为指纹信息来源。首先，针对井下环境复杂存在多种噪声，提出一种指纹构造法，分别获得幅值差和重构相位进行交叉组合，使用UWB技术快速获取指纹相关标签。其次，对多元核极限学习机进行改进，采用分段式量子粒子群算法为模型寻找最优参数，以提高定位精度和泛化性能，并引入在线增量学习和遗忘机制，添加部分新增数据持续更新定位模型，并设置数据有效期遗忘过旧数据减少不良影响。最后将模型输出结果与标签库进行快速匹配，构建完整的指纹定位系统。实验验证，定位误差1m时的置信度为92%，静态平均定位误差为0.39m，动态平均定位误差为0.46m，且受环境多径和遮挡影响较小，拥有更好的长期稳定性。

（3）针对单一传感器定位技术无法满足井下各种环境下高精度定位的问题，采用UWB与Wi-Fi融合定位方案。首先提出一种组合定位设计，在定位区域内部署两个UWB基站和一个AP基站，两者相互辅助获得各自定位坐标并以加权方式组合，然后构建UWB与Wi-Fi融合惯性导航辅助的扩展卡尔曼滤波定位模型，引入自适应因子和抗差因子，自适应的调整预测值和观测值所占的权重，减少观测粗差影响。视距环境下，相对于单一UWB和Wi-Fi定位技术提高了35.7%、52.6%；非视距环境下比传统扩展卡尔曼滤波定位精度提高了28.5%，在精度、稳定性和成本上均优于单一传感器定位技术，该方法能满足煤矿人员定位的精度与稳定性要求。

Coal mine underground personnel positioning system is of great significance to the safe production of coal mines, and the underground tunnel is mostly narrow and long and the environment is complicated and changeable. A single positioning method can not meet the positioning needs under various conditions, and the cost of deploying new equipment is high, in which Wi-Fi with the development of mining wireless communication system gradually achieve full coverage of underground without new equipment. For the special environment of underground, we carry out the research on the application of ultra-wideband UWB ranging and Wi-Fi fingerprinting on underground personnel positioning, and propose a multifuntsion positioning method supplemented by inertial sensors. The main contents are as follows:

(1)Aiming at the clock drift of UWB equipment and the complexity of the underground environment, the traditional TWR dual-range ranging method is improved to reduce the clock drift error, and the multi-base station cooperative communication method is adopted to reduce the number of communication times and shorten the positioning time, while the environmental bias coefficients are added, and the tag ranging results are preliminarily corrected by using the ranging situation between base stations. Then for the multi-state mode of signal transmission during personnel movement, Gaussian mixture model GMM is established to decompose the ranging information, identify the ranging results under line-of-sight conditions and reconstruct them to get the closest real distance information. Finally, the improved coordinate solving algorithm is proposed, and the two-stage maximum likelihood method is designed to estimate the position coordinates. The experimental results show that the ranging accuracy is improved by 68.5%, and the static average positioning error under line-of-sight conditions is 0.28m, which is better than other positioning algorithms in terms of positioning accuracy and stability, and it provides a basis for the rapid collection of fingerprint tags.

(2)Underground Wi-Fi fingerprinting and localisation application exists the problems of sparse distribution of base stations and sparse feature degree, and adopts channel state information CSI as the source of fingerprinting information. Firstly, a fingerprint construction method is proposed for the existence of multiple noises in the complex downhole environment, which obtains the amplitude difference and reconstructed phase for cross-combination, respectively, and uses UWB technology to obtain the fingerprint-related labels quickly. Secondly, the multivariate kernel limit learning machine is improved, and a segmented quantum particle swarm algorithm is used to find the optimal parameters for the model in order to improve the positioning accuracy and generalisation performance, and an online incremental learning and forgetting mechanism is introduced to add part of the new data to continually update the positioning model, and the data validity period is set to forget the old data to reduce the adverse effects. Finally, the model output is quickly matched with the tag library to construct a complete fingerprint localisation system. The experiment verifies that the confidence level is 92% when the positioning error is 1m, the static average positioning error is 0.39m, the dynamic average positioning error is 0.46m, and it is less affected by the environmental multipath and occlusion, and has better long-term stability.

(3)Aiming at the problem that a single sensor positioning technology cannot meet the high-precision positioning in various environments downhole, a UWB and Wi-Fi fusion positioning scheme is adopted. Firstly, a combined positioning design is proposed, where two UWB base stations and one AP base station are deployed in the positioning area, and both of them assist each other in obtaining their respective positioning coordinates and combining them in a weighted way, and then an extended Kalman filter positioning model assisted by UWB and Wi-Fi fusion inertial navigation is constructed, and an adaptive factor and anti-differential factor are introduced, so that the weights of the predicted value and the observed value are adjusted adaptively, and the effect of coarse difference of the observations is reduced. The effect of observation coarseness is reduced. Under line-of-sight environment, it improves 35.7% and 52.6% compared with single UWB and Wi-Fi positioning technology; under non-line-of-sight environment, it improves 28.5% compared with the traditional extended Kalman filter positioning accuracy, which is better than single-sensor positioning technology in terms of accuracy, stability and cost, and this method can meet the requirements of accuracy and stability for positioning of personnel in coal mines.

[1]马丽梅,史丹,裴庆冰.中国能源低碳转型(2015—2050):可再生能源发展与可行路径[J].中国人口·资源与环境,2018,28(02):8-18.

[2]李辉,庞博,朱法华,等.碳减排背景下我国与世界主要能源消费国能源消费结构与模式对比[J].环境科学, 2022, 43(11):5294-5304.

[3]刘晓龙,葛琴,姜玲玲,等.中国煤炭消费总量控制路径的思考[J].中国人口·资源与环境,2019,29(10):160-166.

[4]付恩三,白润才,刘光伟,等.“十三五”期间我国煤矿事故特征及演变趋势分析[J].中国安全科学学报, 2022,32(12):88-94.

[5]唐飞,王云刚,杜炳成,等.基于优化马尔可夫模型的煤矿事故死亡人数预测[J].中国安全科学学报, 2022,32(04):122-128.

[6]蔺瑜,谢贤平,刘杰.矿山安全避险“六大系统”能力成熟度分析[J].中国安全生产科学技术,2023,19(12):59-70.

[7]陆妍玲,韦俊伶,刘采玮,等.融合地磁/WiFi/PDR的自适应粒子滤波室内定位[J].测绘通报, 2020, 37(06):1-6.

[8]李瑶,王健,张涛,等.一种纬度未知条件下捷联惯导抗扰动自对准算法[J].中国惯性技术学报,2020, 28(03):330-337.

[9]李德媛,刑素堂,周林春.煤矿井下人员定位系统技术现状及发展趋势[J].中国高新科技,2021,99(15):62-63+65.

[10]刘送永,崔玉明.煤矿井下定位导航技术研究进展[J].矿业研究与开发,2019,39(07):114-120.

[11]徐淑萍,王双,郭宇,等.一种抗非视距误差的组合定位算法[J].科学技术与工程,2021, 21(31):13405-13412.

[12]张义磊,张兆东.基于哈希函数的矿用RFID定位系统研究[J].中国仪器仪表,2017, (01):54-58．

[13]付波,毛嫚嫚,袁旺军,等.基于RFID定位技术的室内人员行为分析系统[J].湖北工业大学学报,2019,34(02):1-4.

[14]Yang W,Liusheng H,Wei Y,et al. A novel real-time coal miner localization and tracking system based on self-organized sensor networks[J]. EURASIP Journal on Wireless Communications,2010,2010 (1):1-14.

[15]Zhigao L,Chunwen L,Danchen W,et al. A wireless sensor network based personnel positioning scheme in coal mines with blind areas[J]. Sensors, 2010,10(11):9891-9918.

[16]Moridi M A,Sharifzadeh M,Kawamura Y, et al. Development of Wireless Sensor Networks for Underground Communication and Monitoring Systems (the Cases of Underground Mine Environments)[J]. Tunnelling and Underground Space Technology,2018,73(2018): 127-138.

[17]Sun J, Gu R,Li S, et al.HUID: DBN-Based Fingerprint Localization and Tracking System with Hybrid UWB and IMU[J]. China Communications,2023,20(02):139-154.

[18]Khodjaev J,Park Y,Malik A S. Survey of NLOS identification and error mitigation problems in UWB-based positioning algorithms for dense environments[J]. Annals of Telecommunications-Annals Des Telecommunications,2010,65(06):301-311.

[19]罗朋,胡振峰,田世伟,等.UWB雷达芯片的研究现状与发展[J].电子与信息学报,2022, 44(04): 1176-1192.

[20]刘万青,魏国,高春峰,等.基于深度学习的UWB NLOS传播影响抑制技术研究[J].红外与激光工程,2023,52(12):293-301.

[21]鞠晨.矿井UWB定位系统中圆极化天线的优化设计和应用[J].工矿自动化,2023, 49(01):171-176.

[22]尹秋园,余慧敏.TDOA/SDS-TWR联合的超宽带室内测距[J].电子测量技术,2021, 44(04): 75-79.

[23]邵小强,李康乐,陈熙,等.基于改进卡尔曼滤波和参数拟合的矿井TOA定位方法[J].煤炭学报,2019,44(05):1616-1624.

[24]Voronov R,Moschevikin A,Soloviev A. Algorithm for Smoothed Tracking in Underground Local Positioning System[C]// 2018 International Russian Automation Conference. IEEE,2018:1-6.

[25]Li K,Wang H,Li S. A mobile node localization algorithm based on an overlapping self-

adjustment mechanism[J]. Information Sciences, 2019, 481: 635-649.

[26]Guo X,Xia P. Mine personnel location system based on Internet of things[C]// IOP Conference Series: Materials Science and Engineering.IOP Publishing,2018,439(3):32-41.

[27]宋明智.矿井WLAN位置指纹定位中的关键问题研究[D].中国矿业大学,2020:4-13.

[28]孙哲星,孙继平.异步测时矿井人员精确定位方法[J].煤炭学报,2018,43(05):1464-1470．

[29]宋明智,钱建生,胡青松.基于AKPCA算法的井下WLAN位置指纹定位[J].煤炭科学技术,2022,50(03):248-256.

[30]吴静然,崔冉,赵志凯,等.矿井人员融合定位系统[J].工矿自动化,2018,44(04): 74-79.

[31]马京.基于指纹膜与航迹推算的煤矿井下人员定位技术研究[D].中国矿业大学,2017.

[32]康瑞清.建筑物内复杂环境下的地磁场定位导航研究[D].北京科技大学,2016.

[33]Zheng Yang,Zimu Zhou,Yunhao Liu. From RSSI to CSI: Indoor Localization via Channel Response[J]. ACM com-puting surveys, 2014,46(2):25.1-25.32.

[34]张大庆,张扶桑,吴丹,等.基于CSI的通信感知一体化设计：问题、挑战和展望[J].移动通信,2022,46(05):9-16.

[35]王旭东,刘帅,吴楠.CAEFi:基于卷积自编码器降维的信道状态信息指纹室内定位方法[J].电子与信息学报,2022,44(08):2757-2766.

[36]Q. Li, X. Liao, M. Liu, et al. Indoor localization based on CSI fin-gerprint by siamese convolution neural network[J]. IEEE Trans. Veh.Technol.,2021,70(11):12168-12173.

[37]张雷.基于CSI的煤矿井下定位方法研究[D].中国矿业大学,2019:15-64.

[38]Zhang L,Hu Y,Liu Y,et al. WiCLoc: A Novel CSI-based Fin-gerprint Localization System[J]. International Journal of Pattern Recognition and Artificial Intelligence, 2020, 34(9): 1-12.

[39]王国法,赵国瑞,任怀伟.智慧煤矿与智能化开采关键核心技术分析[J].煤炭学报,2019, 44(01):34-41.

[40]Xu Y, Li Y Y, Ahn C K, et al. Seamless indoor pedestrian tracking by fusing INS and UWB measurements via LS-SVM assisted UFIR filter[J]. Neurocomputing,2020,38(08):301-308.

[41]Li Y,Zhuang Y,Zhang P,et al. An improved inertial/wifi/magnetic fusion structure for indoor navigation[J]. Information Fusion, 2017, 34:101-119.

[42]吴怡,李希胜.融合PDR/地磁指纹的室内定位算法[J].传感器与微系统,2022,41(03): 132-134.

[43]陈国良,张言哲,汪云甲,等.WiFi-PDR室内组合定位的无迹卡尔曼滤波算法[J].测绘学报, 2015, 44(12): 1314-1321.

[44]吴静然,崔冉,赵志凯,等.矿井人员融合定位系统[J].工矿自动化, 2018, 44(04): 74-79.

[45]宰昶丰,郭英,叶瑾,等.基于UWB穿墙定位中NLOS误差的优化方法[J].电子测量技术, 2021, 44(19):128-133.

[46]孟水仙,任海军,宋岩熠,等.UWB信号与5G设备同频干扰研究[J].无线电工程,2022, 52(08):1354-1360.

[47]刘正波, 朱亮.LOS/NLOS混合环境下的基于TOA测距的定位算法[J].弹箭与制导学报, 2021, 41(02):78-81.

[48]孟天次,张贞凯,林云航.一种改进两步加权最小二乘的TDOA定位算法[J].电讯技术, 2022, 62(06):782-787.

[49]谢雪,王浩祥,邓平.一种基于单次散射体定位的TOA/AOA混合定位算法[J].电讯技术,2022,62(01):110-115.

[50]邹东尧,陈鹏伟,刘宽.一种改进的RSSI测距定位算法[J].电讯技术,2019,59(10): 1191-1196.

[51]王飞.基于UWB TOF测距的矿井精确定位方法研究[J].自动化仪表, 2022, 43(08):62-66.

[52]Li Shaoxian,Fan Lii,Wang Chengrui, et al. UWB positioning enhancement using Markov chain in indoor NLOS environment [J]. The Journal of China Universities of Posts and Telecommunications, 2020, 27(4):6.

[53]Halperin D,Hu W,Sheth A, et al. Predictable 802.11 packet delivery from wireless channel measurements[J]. ACM SIGCOMM Computer Communication Review, 2011, 41(4): 159-170.

[54]Wang J,Zhang L,Wang X,et al. A novel CSI pre-processing scheme for device-free localization indoors[C]// Proceedings of the Eighth Wireless of the Students, by the Students, and for the Students Workshop.ACM,2016:6-8.

[55]周燕,李华旺,张永合.基于有界网格的贝叶斯滤波位置估计算法[J].激光与光电子学进展,2022,59(05):108-115.

[56]周楚霖,陈敬东,黄凡.基于无迹粒子滤波的WiFi-PDR融合室内定位技术[J].计算机科学, 2022, 49(S1): 606-611.

[57]Guo S L, Sun Y J, Chang L M, et al. Robust cubature Kalman filter method for the nonlinear alignment of SINS[J].Defence Technology,2021,17(02):593-598.

[58]黄小平,王岩.卡尔曼滤波原理及应用[M].北京:电子工业出版社, 2015: 30-32.

[59]王傲明,李姗姗,李新星,等.基于自适应并行扩展卡尔曼滤波的SITAN匹配算法[J].中国惯性技术学报,2022,30(01):81-88.

[60]Tian X H,Li W X,Yang Y C,et al. Optimization of Fingerprints Reporting Strategy for WLAN Indoor Localization[J]. IEEE Transactions on Mobile Computing,2018,17(2):390-403.

[61]Cheng Hua,Kun Zhao,Danan Dong,et al. Multipath Map Method for TDOA Based Indoor Reverse Positioning System with Improved Chan-Taylor Algorithm[J]. Sensors,2020, 20(11):3223-3223.

[62]胡荣明,苏瑞鹏,竞霞,等.融合改进小波去噪与T-Taylor的井下定位算法[J].测绘通报,2023,No.551(02):46-51.

[63]刘星,王文双,赵建印,等.自适应在线增量ELM的故障诊断模型研究[J].系统工程与电子技术,2021,43(9):2678-2687.

[64]Weipeng C,Ming Zhong X,Zhiwu ZJ, et al. Online sequential extreme learning machine with dynamic forgetting factor[J]. IEEE Access,2019,7:179746–179757.

[65]张兆娟,王万良,唐继军.适应度二次选择的QPSO和SA协同搜索大规模离散优化算法[J].通信学报,2020,41(8):22-31.

[66]Wang Xuyu,Wang Xiangyu,Mao Shiwen. CiFi:Deep convolutional neural networks for indoor localization with 5 GHz Wi-Fi[C].Proceedings of 2017 IEEE International Conference on Communications (ICC),Paris,France, 2017:1–6.

[67]Bi S,Ma L,Shen T,et al. Neural network assisted Kalman filter for INS/UWB integrated seamless quadrotor localization[J].PeerJ Computer Science,2021,7(17):e630.

[68]Jiang C,Zhang S,Li H,et al. Performance evaluation of the filters with adaptive factor and fading factor for GNSS/INS integrated systems[J]. GPS Solutions,2021,25(4):130-140.