基于信道状态信息的煤矿井下协同定位方法研究

The research on underground coal mine target localization has great practical significance for production safety. The traditional localization methods, such as received signal strength-based, time-based, and fingerprint-based localization methods, cannot be used in underground mines directly due to the complicated environment and serious multi-path effects. Hence, the problem of underground target localization calls for further discussions and studies. In this proposal, channel state information is collected by common WiFi devices, and its effective features are extracted according to the requirements of active and passive localization. On this basis, the active, passive and cooperative localization methods are studied step by step. The main research contents include: (1) The decay mechanism is studied for underground wireless communication channels. And then the on-demand feature extraction method of channel state information is introduced. (2) The adaptive ranging and direction-finding models are built based on the extracted features. Furthermore, the active mine localization approach is proposed by using communication devices carried by the target. (3) To solve the abnormality or failure problems of underground communication devices, the target position mapping model is built in the light of the channel state information. Moreover, the passive device-free localization method is presented for coal mine. (4) Aiming at the restricted problem between the energy and precision, the fuzzy decision-making strategy is given which can fuse the active and passive target localization methods. The energy efficient cooperative localization method is then obtained for underground mine targets. Based on the above systematic researches, series of theoretical results will be obtained. The results obtained will provide a new idea for underground coal mine target localization as well as lay the foundation for the practical application.

矿井定位对于保障煤矿安全生产具有重大现实意义。井下环境恶劣、多径严重，传统基于信号强度、时间及指纹等定位方法难以直接应用到井下，煤矿井下定位问题有待深入探索和研究。本课题利用WiFi设备采集信道状态信息并按需提取面向主被动定位应用的有效特征，在此基础上依次递进研究井下目标主动定位、被动定位及协同定位方法。具体包括：(1)探讨井下信道状态衰变机理，研究信道状态信息多维特征和变化特征的按需提取方法。(2)建立基于信道状态多维特征的环境自适应测距与测向模型，研究基于设备的井下目标主动定位策略。(3)针对井下定位终端异常失效问题，构建基于信道状态变化特征的目标位置映射模型，研究无需设备的井下目标被动定位策略。(4)针对井下能量与精度相互制约问题，深度融合主被动定位策略，研究能效均衡的井下目标协同定位方法。本项目的研究将得到系列性理论成果，为煤矿井下定位提供新思路，同时对于实际应用也有重大工程价值。

煤炭是我国战略支柱能源，未来煤炭仍是我国能源体系的压舱石和兜底保障。井下定位技术是实现煤矿智能化的重要手段，然而煤矿井下环境恶劣、多径严重，传统基于信号接收强度的定位方法难以直接应用到井下。为此，本项目主要研究了基于物理层信道状态信息的定位方法，包括：（1）Wifi信道状态信息特性研究。结合井下生产环境实际需求，搭建了基于迷你工控机、笔记本及智能手机的三类信道状态信息提取平台，分析了信道状态信息振幅和相位特性，探讨了振幅与距离、相位与感知的关系。（2）基于信道状态信息的主动定位方法研究。针对指纹库构建复杂、迁移难的问题，本项目通过信道状态信息的三种域表示方法，设计了基于迁移学习的指纹定位方法；针对现有到达角算法依赖天线数量与相位校准的问题，提出了无需相位校准的测向方法。（3）基于信道状态信息的被动感知方法研究。设计了基于信道状态信息的被动式人体活动感知方法，实现大尺度和小尺度活动识别；为进一步提高感知方法的可迁移性，以适用新的数据域，设计了基于对抗生成网络的跨域识别方法。（4）煤矿井下协同定位方法研究。基于信道状态信息设计协同定位框架，融合主被动定位结果，提高了定位精度；在考虑目标定位精度与能量耗费的基础上，提出多传感器融合定位方法，实现能效均衡的井下目标按需定位。经过四年的研究，本课题取得了系列性研究成果，对煤矿井下定位有重要的理论价值与实际意义。