基于矿区手机移动传感网络的矿震时空强测算方法研究

With the mining depth of coal resources increasing, China has become one of the most serious countries threatened by the mine earthquake. Due to the lack of a reliable and practical mine earthquake warning system, the destructive mine earthquake accidents still occur occasionally. With the rapid development of mobile Internet, the operating system with Android and IOS for mainstream makes smart phone have the same scalability and a strong sense of intelligence as the PC system. In view of the defects of existing mine seismic monitoring system, such as high cost, low number of stations and low positioning accuracy. The foundation item will make full use of the high density and mobility of mining workers’ mobile phones, client software of mobile sensor networks is designed for mine earthquake monitoring. It will be used to identify mine earthquake signal and jamming signal of human activities. We will design measuring and calculating method for mine earthquake space time magnitude based on mobile sensor networks based on mobile sensor networks trigger algorithm model under noisy environment. When a trigger event is recognized as mine earthquake, according to the information of the trigger mobile phone in the network, the method will update the mine earthquake origin time, earthquake source position and magnitude continuously. It will improve the measuring and calculating precision of mine earthquake space time magnitude by optimizing algorithm continuously in the field application.

随着煤炭资源开采深度的增加，我国已成为受矿震威胁最为严重的国家之一，由于缺乏可靠实用的矿震预警系统，破坏性矿震事故时有发生。伴随移动互联网的快速发展，以Android和IOS为主流的操作系统使得智能手机具备如同PC系统一般可扩展性，具备较强的智能“感知”能力。针对现有矿山微震监测系统价格昂贵、台站数量少、定位精度低等缺陷，本项目拟充分利用矿区职工手机的高密度性和移动性，研发手机移动传感网络矿震监测客户端软件。通过识别矿震信号和人体活动干扰信号，建立嘈杂环境下手机移动传感网络触发算法模型，设计手机移动传感网络矿震时空强算法，当产生矿震触发事件时，该算法将根据网络中触发手机信息连续不断更新矿震发震时间、震源位置和震级大小。现场应用不断优化算法，将有效提高矿震时空强测算精度。

项目为提高矿震监测系统定位精度，减少监测盲区，降低监测成本，提出一种基于矿区手机移动传感网络的矿震时空强测算方法。首先开发了手机端数据采集APP，通过实验研究了不同手机间精度、与人类日常活动区别、振动台模拟、煤岩加载破坏手机震动信号监测试验，为实际应用提供了基础；其次提出一种传感器网络实时感知流式数据特征事件实时分阶判识方法，通过确定流式数据事件的判识因子，基于流式数据的漂移特征构建动态时间规整反走样模型，引入可变滑动窗口机制获取事件发生的起始判识域，找到相似匹配的判识比例，提出基于判识比例阈值分配的事件分阶实时判识方法，通过矿震事件的初实时快速判识和终实时精确判识，得到事件发生的可能性以及最终矿震事件的信息；以及建立了以矿区附近工人及家属等使用的智能手机建立矿震手机监测网络，对模拟震源点网格化，构建基于标准差的目标函数，提出改进的萤火虫寻优策略，并使用拐点回溯法以及手机移动传感网络排除离散点策略(EDPS)降低定位误差；最后通过矿震实验进行验证，实验结果表明：在手机移动传感网络无到时误差理想情况下，所有模拟震源点都能够准确收敛至震源位置，定位误差小于1m，当手机相较于检波器到时误差较高，且定位误差与到时误差具有相关性，当手机到时误差为-1~1s时，传统算法定位误差为216m，无法实现高精度定位。通过研究目标函数值与定位误差间的关系，提出并使用拐点回溯法以及EDPS两种优化方法，定位算法绝对定位误差降低至72m，当到时误差为-0.2~0.2s时，绝对定位误差降低至17m，定位精度提高66.7%。通过项目研究证明基于手机移动传感网络监测矿震方法可行，具有重要科学意义和应用价值。