【作者】 刘卫东；

【导师】 丁恩杰； 焦李成；

【作者基本信息】 中国矿业大学 ， 通信与信息系统， 2009， 博士

【摘要】 声发射(Acoustic Emission, AE)技术是无损监测评估中的一种重要方法,目前声发射预测冲击地压的普及仍然存在很多方面的问题,主要存在且急需解决的是抗噪声性能差、定位不准确及预测预报需要人为干预等问题。本文在分析冲击地压机理的基础上,从声发射信号处理的角度对以下关键问题进行研究:1.准确获取能表征煤岩体特征的声发射信号是准确预测冲击地压的前提条件。针对井下噪声复杂多变且具有特殊规律的环境,在分析矿井噪声特性的基础上,提出并设计了双自适应滤波器,并应用于预测冲击地压声发射信号的消噪处理。同时对双自适应滤波器涉及的两个方面:自适应滤波算法和自适应非均匀子带滤波器进行分析设计。2.矿井声发射监测中定位主要采用被动时差定位法,其中时间延迟估计是最关键、最核心的问题,其准确程度直接决定定位的精度。声发射信号在煤岩体中的传播特性直接影响其定位算法的选择,本文首次搭建了声发射信号在煤岩体中传播的物理模型,并以此建立了以数字滤波器级联构成的数学模型。在建立时差定位模型的基础上,提出并设计了小波分析和广义互相关相结合的时延估计方法。3.声发射预测预报专家系统可以弥补矿井人才缺乏,提高预测的精度。本文将在实验室测定的煤岩层冲击倾向性指标和现场的声发射监测数据相结合,采用层次分析法建立冲击地压预测预报的三层模型。并且将自适应模糊神经网络引入模型中来训练隶属度函数,克服由于人为原因造成预测的误差。通过仿真证明了方法的可行性和准确性。4.传感器的布置一方面影响时差定位的准确性,另一方面影响系统成本。在矿井中对声发射传感器的布置与其他方面的应用很大的不同的点就是随着开采区域的扩展,需要监控的区域不断扩大。如何通过增加少数传感器可以达到最佳的监控效果是必须要考虑的问题。同时为了弥补微弱声发射信号只被一个固定传感器检测到,导致无法准确定位的缺点,提出了一种无线传感器网络与有线网络相结合的监测矿震的方法,建立监控模型,并对无线网络的定位算法进行了分析和改进。更多还原

【Abstract】 AE (Acoustic Emission, AE) is an important method in nondestructive monitoring evaluation, but there are still many problems in the popularity of AE forecasting rockburst. The main problems to be resolved are poor anti-noise performance, inaccurate location and forecasting which requires human intervention etc. Based on the analysis of the impact of the mechanism,this paper studys the following key problems from the perspective of AE signal processing.1. To obtain accurate AE signals which can represent mass characteristics of coal and rock is prerequisite to accurately predict the rockburst. The noise sources underground are multitudinous and complex, and have an environment with special rules. Based on the analysis of interference noise characteristics, the paper proposed to design a pair of adaptive filters to counteract the noise signal in ground pressure on the AE signal. The double adaptive filter involves two aspects: the adaptive filtering algorithms and adaptive non-uniform subband filter.2. Passive TDOA location method is mainly adapted in monitoring mine AE location and time delay is estimated to be the most crucial core question. The characteristics of AE signal propagation in coal and rock directly affect its localization algorithm. This paper constructs the physical model of the AE signal transmission in coal and rock for the first time, and establish mathematical model. Based on analysis of noise propagation characteristics and distribution, the paper proposes and designs time delay estimation method which combines with the wavelet analysis and the cross-correlation method.3. AE forecasting expert system can compensate for the shortage of qualified personnel of mine, improve forecast accuracy. The paper establishes the three layers model of rock prediction, combining with coal strata impact tendentiousness index measured in the laboratory and the AE monitoring data measured in the mine, and introduces the adaptive fuzzy neural network to train membership functions, overcomes the prediction error caused by individual. The feasibility and accuracy of this method has proved through data simulation.4. The layout of the sensor affects location accuracy of AE source positioning on one hand, on the other hand directly impacts on the system cost. The very different point between the arrangement of AE sensors in mines and other applications is that as the expansion of mining areas is more and more covered, the areas of being monitored is increasing as well, it must be taken into account that how to reach the best monitoring effect sensors adding a few sensor. Week signals were only detected by a fixed sensor which failed in accurate locations. In order to overcome this shortcoming, this paper proposes a wireless sensor network to complement the wired network, establishes the monitoring method for wireless sensor network monitoring rock orientation, analyzes and improves the positioning algorithm of wireless networks.更多还原