两级数据融合算法在煤矿粉尘监测中的应用研究

发布时间：2018-03-01 20:15

  本文关键词： 数据融合 煤矿粉尘 实时监测 D-S证据理论 RS理论　出处：《太原理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着煤矿生产自动化深入推进,矿井产尘呈现出量大繁杂、管控难的问题。作为目前较为先进的降尘装置,煤矿粉尘在线监测系统虽然实现了粉尘浓度实时连续监测,但由于监测指标单一、数据信息冗错高的缺陷导致其决策失真、误动作频发,严重影响到除尘效率,同时也造成水资源极大浪费。因此,本文将两级数据融合算法应用于煤矿粉尘监测系统。利用数据层(第一级)融合算法对同质传感器原始数据进行过滤剔除和消除错误冗余信息,并在分析基于D-S证据理论的算法和基于RS理论的算法这两种决策层融合算法单一应用不足的基础上,提出了综合D-S证据理论和RS理论的决策层新算法。从理论分析的角度证明了算法可行性,并开展了不同工矿位置的现场应用。主要研究内容概括如下:(1)在分析煤矿粉尘在线监测系统中传感器监测指标选取和数据处理的基础上,将传感器监测指标丰富细化为全尘浓度、呼尘浓度、粉尘粒度、风速四项,确保了除尘决策的全面有效。建立煤矿粉尘监测数据库,为两级数据融合算法在煤矿粉尘监测系统中的应用奠定了基础。(2)为适应矿井环境复杂、工况多变的监测环境,建立了数据融合算法的两级结构模型,明确了算法的层次和地位。数据层融合选取基于支持度矩阵的算法,在分析阐明算法主要思想的同时概括提炼出清晰的算法步骤,并以粉尘监测数据库数据为对象进行了算例验证。在分析基于D-S证据理论的算法和基于RS理论的算法这两种决策层融合算法的前提下,通过煤矿粉尘监测中的应用算例对两种算法的决策结果进行了对比,发现这两种算法决策规则单一等问题往往导致决策结果不够准确,可信度不高。(3)提出了综合D-S证据理论和RS理论的决策层新算法,从理论分析的角度证明了算法可行性,针对新算法实现过程中的三个关键问题(BPA函数获取、属性重要度评估、证据合成方法)做了重点分析,进一步完善了新算法的理论基础和运算规则。最后,以相同监测信息的处理为例证明了新算法比单一基于D-S证据理论的算法或基于RS理论的算法更具优势。它不仅可以客观确定BPA函数,以概率形式输出决策结果,而且在粉尘决策精确度和可信度方面均有明显提升。(4)对两级数据融合算法在煤矿粉尘监测中的应用进行了实例验证,然后在试验煤矿中以粉尘控制和节水效果为参考,研究测试了两级数据融合算法对煤矿粉尘监测系统的改进效果。结果表明,在确保粉尘控制效果的前提下,改进后系统喷雾装置工作时间较改进前平均减少了30%左右,喷水量平均下降了20%左右。现场应用结果表明数据融合算法的应用对粉尘在线监测系统的功能和效率有了明显改善提升,验证了本文所提方法是正确可行的。
[Abstract]:With the further development of coal mine production automation, the problem of large and complicated quantity and difficult control of mine dust production appears. As a relatively advanced dust control device, the on-line monitoring system of coal mine dust has realized the real-time and continuous monitoring of dust concentration. However, the defects of single monitoring index and high redundancy of data information lead to the decision distortion and frequent misoperation, which seriously affect the efficiency of dust removal, and also cause a great waste of water resources. In this paper, the two-level data fusion algorithm is applied to the coal mine dust monitoring system. The data layer (first level) fusion algorithm is used to filter and remove the original data of the homogeneous sensor and eliminate the error redundancy information. Based on the analysis of the single application of the two decision level fusion algorithms based on D-S evidence theory and RS theory, A new decision layer algorithm based on D-S evidence theory and RS theory is proposed. The feasibility of the algorithm is proved from the point of view of theoretical analysis. The main research contents are summarized as follows: (1) on the basis of analyzing the selection of sensor monitoring indexes and data processing in online monitoring system of coal mine dust, The rich monitoring index of sensor is divided into four items: total dust concentration, exhaling dust concentration, dust particle size and wind speed, which ensures the overall effectiveness of dust removal decision. The coal mine dust monitoring database is established. For the application of two-level data fusion algorithm in coal mine dust monitoring system, a two-level structure model of data fusion algorithm is established to adapt to the complex and changeable monitoring environment of mine environment. The level and position of the algorithm are defined. The data layer fusion selects the algorithm based on support matrix, and summarizes the clear algorithm steps while analyzing and clarifying the main ideas of the algorithm. Taking the data of dust monitoring database as an example, the paper analyzes the two decision level fusion algorithms based on D-S evidence theory and RS theory. The decision results of the two algorithms are compared through the application examples in coal mine dust monitoring. It is found that the decision results of the two algorithms are often inaccurate due to the single decision rules of the two algorithms. In this paper, a new decision layer algorithm based on D-S evidence theory and RS theory is proposed. The feasibility of the algorithm is proved from the point of view of theoretical analysis. In view of the three key problems in the implementation of the new algorithm, BPA function acquisition and attribute importance evaluation are discussed. The method of evidence synthesis) makes the emphasis analysis, further consummates the new algorithm the theory foundation and the operation rule. Finally, Taking the processing of the same monitoring information as an example, it is proved that the new algorithm is superior to the single one based on D-S evidence theory or RS theory. It can not only objectively determine the BPA function, but also output the decision result in the form of probability. The application of two-level data fusion algorithm in coal mine dust monitoring is verified by an example, and then the dust control and water-saving effect are used as references in the experimental coal mine. The improvement effect of two-stage data fusion algorithm on coal mine dust monitoring system is studied and tested. The results show that the working time of the improved spray system is about 30% less than that before the improvement, on the premise of ensuring the dust control effect. The field application results show that the application of the data fusion algorithm has improved the function and efficiency of the dust on-line monitoring system obviously, and verified that the method proposed in this paper is correct and feasible.
【学位授予单位】：太原理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TD714.3;TP202

【参考文献】

中国期刊全文数据库 前10条

1 陈曦;葛少成;徐耀松;樊文涛;葛斐;;大型选煤厂粉尘在线监测与智能喷雾降尘系统研究[J];安全与环境学报;2015年05期

2 郝叶军;;光散射性矿用粉尘传感器的设计及应用[J];煤矿机电;2015年05期

3 何国家;杨利峰;龙如银;祁慧;陈红;;煤矿粉尘危害风险规模评价模型[J];中国煤炭;2015年08期

4 肖丹;;煤矿井下粉尘超标原因分析及其综合管理探讨[J];煤矿现代化;2015年04期

5 郑雷;王雪芹;;煤矿粉尘爆炸事故树分析[J];中国矿山工程;2015年03期

6 姜家兴;王德明;;矿尘浓度监测技术现阶段存在的问题及改进措施[J];煤炭技术;2015年01期

7 张久红;邵春华;葛宪民;;研究探讨尘肺病的综合预防与治疗[J];职业卫生与病伤;2014年05期

8 许满贵;刘欣凯;文新强;何鹏程;朱兆伟;;煤矿综采工作面粉尘分布及运移规律研究[J];西安科技大学学报;2014年05期

9 周ml皋;;呼吸性粉尘分离效能测试方法误差分析[J];矿业安全与环保;2013年06期

10 李琼;杨晓祥;;多传感器数据融合模型的研究[J];计算机与现代化;2013年04期

中国博士学位论文全文数据库 前2条

1 韩磊;华东某煤矿粉尘污染变化趋势、煤工尘肺发病特征及其预测研究[D];南京医科大学;2016年

2 崔波;多传感器目标跟踪数据融合关键技术研究[D];西南交通大学;2012年

中国硕士学位论文全文数据库 前10条

1 柴倬君;双Pt矿用风速传感器的研究[D];东华理工大学;2016年

2 李凤娟;数据融合技术在基于物联网的火灾探测系统中的应用研究[D];吉林大学;2015年

3 王力;基于DS证据理论的多传感器数据融合算法研究与应用[D];太原理工大学;2015年

4 解春香;异类多传感器数据融合技术的研究[D];沈阳理工大学;2015年

5 樊文涛;塔山选煤厂粉尘在线监测与治理技术研究[D];辽宁工程技术大学;2015年

6 刘小贝;粉尘微粒监测系统的研究与设计[D];长春工业大学;2014年

7 敖蕾蕾;基于D-S证据理论的电网设备状态检修决策研究[D];浙江大学;2014年

8 张丽;中国主要城市空气质量与经济发展的关系研究[D];华北水利水电大学;2013年

9 石西庆;基于数据融合技术的电子政务信息共享服务平台模型[D];电子科技大学;2013年

10 胡雅馨;基于粗糙集与证据理论的瓦斯传感器故障诊断技术的研究[D];辽宁工程技术大学;2009年

，

本文编号：1553413