煤矿物联网多源异构监测信息源端压缩采集方法研究

项目名称： 煤矿物联网多源异构监测信息源端压缩采集方法研究

项目编号： No.51204186

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 冶金与矿业学科

项目作者： 徐永刚

作者单位： 中国矿业大学

项目金额： 25万元

中文摘要： 物联网技术在开放空间已被大量应用."感知矿山物联网"试图将物联网技术引入地下,实现对井下受限空间中特定对象多维特征的实时监控,其研究尚处于起步阶段.矿山物联网多源异构传感信息的汇集使单位信道所需传输带宽激增,对矿山井下现有通信线路,尤其自传感源至环网信道带来巨大压力,以电缆为主的现有线路成为制约矿山物联网应用的瓶颈.线路改造虽能解决瓶颈问题,但耗费巨大. 本研究首次将压缩感知理论及井下空间信息引入矿山安全监控信息采集中,充分利用监控信息的互相关性和可稀疏性,实现将信号的采集转变为对信息的采集,大大降低数据采集量,降低对传输线路的带宽要求.课题通过对(1)矿山安全生产监控数据的特征分析及稀疏化表示，建立正交基数据字典(2)结合专家知识构建观测矩阵(3)稀疏矩阵的凸优化求解(4)信号的重建与评估四方面内容的研究,提出适合于煤矿的压缩感知理论相关算法,为突破矿山物联网传输瓶颈提供有益的理论支持.

中文关键词： 矿山物联网；压缩感知；多尺度1-bit压缩感知；自适应观测矩阵；残差收敛正交匹配追踪

英文摘要： Internet of Things technology has been widely used in open space."Perception mine IoT"tries to introduce the IoT technology into mine underground Intranet,to capture the real-time multi-dimensional characteristics of a monitored object from the limited space,however,the research is just in its beginning.The large number of data collected by those sensors makes the bandwidth required rapid increasing in mine IoT,which makes tremendous pressure on the current communication channel, especially between the channel of source-sensors to fiber looped network,the cable-based channel become the bottleneck of restricting the mine IoT development while line updating expensive cost. This research first introduces compressive sensing theory and underground space location into information collection in the mine safety monitoring ereas,which makes full use of the cross-correlation and sparsity to transform signal sampling into information collection to reduce the massive sampling data,with the result,to release the pressure of the cable-based channel.The issue includes:(1)Analysising characteristic and sparsing representation on the monitored data of safety in mine,establishing orthogonal basis dictionary(2)Building observation matrix combined with expert knowledge(3)Solving the sparse matrix convex optimization(4) reconstru

英文关键词： Mine of Things (MoT)；Compressive sensing(CS)；Multi-scale 1-bit CS；Adaptive observation matrix；Residual Convergence Orthogonal matching pursuit