【作者】 王红尧；

【导师】 徐钊； 华钢；

【作者基本信息】 中国矿业大学 ， 通信与信息系统， 2009， 博士

【摘要】 本文以安徽省淮北矿业集团资助项目—“网络化钢丝绳损伤在线快速检测系统”为背景,根据矿用提升钢丝绳的具体使用环境,选择漏磁检测技术作为其损伤的在线检测手段。针对目前漏磁检测方法的特点、检测能力以及钢丝绳在线检测中存在的问题,进行理论、仿真和实验研究,论文主要工作如下:(1)建立了钢丝绳缺陷漏磁信号空间分布分析模型,对漏磁信号的形成机理和漏磁特性进行了研究,发现钢丝绳缺陷空间漏磁场主要集中分布在钢丝绳轴向方向,径向方向漏磁很少,研究了钢丝绳缺陷几何尺寸和提离值对漏磁检测信号的影响,得出了具有重要参考价值的影响规律,为钢丝绳探伤传感器的优化设计提供了理论指导。通过实验方法,对本文建立的漏磁信号空间分布分析模型进行了有效性验证。(2)提出了煤矿提升钢丝绳在线检测系统结构模型,建立了钢丝绳磁化回路结构分析模型,得出了磁化装置结构参数对钢丝绳局部磁化效果的影响规律,并在相关结论的基础上,将有限元分析方法与优化方法相结合对钢丝绳磁化回路结构主要参数进行了优化设计。(3)提出了一种基于聚磁原理的多霍尔元件阵列组合检测方法,该方法只需8片高灵敏度集成霍尔元件便可实现大直径钢丝绳的全方位无漏检测,且对钢丝绳绳股漏磁场、背景磁场的干扰有较好的抑制效果,该方法从检测信号波形上明显区分开LF型损伤和LMA型损伤,实现了LF、LMA两种类型损伤的同时检测,比原有检测方法前进了一大步。(4)建立了钢丝绳漏磁检测信号信息处理模型,搭建了钢丝绳在线检测实验平台,对钢丝绳损伤进行在线检测,针对真实的钢丝绳漏磁在线检测信号,提出通过信号相关性法选取最优小波基函数和最优小波分解层数的方法,得出最优小波基函数是sym6,最优小波分解层数是4。对本文提出方法的可行性进行了实验验证。针对实际的钢丝绳漏磁在线检测信号,提出基于一种新阈值函数的自适应小波阈值消噪法,仿真实验结果表明,本文提出的方法较目前的小波阈值函数去噪法有较高信噪比和较小最小均方误差,且对钢丝绳小损伤(断丝1-2根,且断口宽度在5mm以内)产生的小奇异信号具有较好的识别效果。(5)对钢丝绳漏磁在线检测信号进行了特征提取,建立了钢丝绳漏磁检测信号RBF神经网络识别模型,运用更有效的IRBF神经网络学习算法对钢丝绳损伤进行定量识别,钢丝绳断丝准确识别率达到96%,定量识别精度较高。更多还原

【Abstract】 This thesis investigates on on-line detection methods of coal mine-hoist wire rope flaws, whose background is the project“Network detection system of coal mine-hoist wire rope”supported by HUAIBEI MINING(GROUP) CO.LTD. In this study, magnetic flux leakage detection technologies are choosen as the method of detecting wire rope flaws aiming at the special application environments of coal mine-hoist wire rope. According to the characteristics, detection ability and defaults in present detection methods of wire rope, some theory, simulation and experimental study have been carried out.This study includes the following contents:(1) The spatial distribution analysis model of magnetic flux leakage signal for wire rope defaults is established, and the formative mechanism and leakage magnetic characteristic of magnetic leakage signal have been studied. The impact of geometrical dimension and lift-off value on magnetic flux leakage signal also have been studied, and obtained influence rule with an important reference value which provides a theoretical guidance for optimizing the design of wire rope flaw detection sensor. The spatial distribution analysis model of magnetic flux leakage signal was validation through experimental methods.(2) The FEM analysis model of magnetization loop structure is built. The impact relation curve of magnetization loop structure parameter on local magnetization effect is educed. And based on this, this study combines FEM and optimization algorithm to have an optimization design of main parameters of magnetization loop structure.(3) Based on the principle of magnetism, we proposed a detection method of multi-hall-element combined with array, which only needs 8 pieces of high sensitivity integrated hall element to implement all-sides detection about large diameter wire rope; what’s more, it has good inhibitory effect to magnetic leakage from wire rope strands and background interference; what’s the most, this method can detect two types of damage (LF and LMA) at the same time.(4) This study established an information processing model of wire rope magnetic flux leakage detection signal, set up an on-line detection experiment platform to detect wire rope damage. Meanwhile, based on the real magnetic flux leakage detection signal, this study proposed the method of selecting the optimal wavelet function and the optimal wavelet decomposition level through the signal correlation,then the results is given that the optimal wavelet basis function is sym6 , the optimal wavelet decomposition level is 4. The feasibility of this method is validated. According to the real magnetic flux leakage detection signal of wire rope, an adaptive wavelet thresholding de-noising method based on a novel thresholding function is proposed. The simulation results show that the method proposed in this paper,compared to the present wavelet threshold functions method, has higher signal to noise ratio and smaller minimum mean-square error, and has a better recognition effect on small singular signals produced by small damage.(5) The feature extraction of magnetic flux leakage signals about wire rope is finished. The quantitative identification using Improved Radial Basis Function neural network model for wire rope damage show that the accuracy of the recognition rate of wire rope damage is up to 96%.This method has higher recognition accuracy.更多还原