齿轮故障信号多重分形维数的形态学计算方法

针对传统的多重分形维数计算方法的缺陷,提出基于数学形态学操作的多重分形维数计算方法,并证明了与盒计数法计算多重分形维数的一致性.对实际的齿轮正常、齿面磨损故障和断齿故障信号进行了分析,结果表明,与盒计数法相比,基于数学形态学计算的多重分形维数能够准确区分齿轮的工作状态,并且数学形态学只涉及简单的加减和取大、取小运算,计算更加简单、快速.     

基于分形维数的GIS局部放电信号发展过程分析

介绍了分形理论中盒维数、信息维数、关联维数的相关定义,分析了其各自的特点,对GIS局部放电加压试验中的固体金属颗粒故障的局放信号进行分维数计算,并阐述了分维数在不同放电阶段的变化,比较了各分维数对时域信号进行描述的特点。     

基于分形理论的摩擦焊超声检测缺陷信号识别

采用超声波探伤仪U1traPAC系统对GH4169高温合金摩擦焊件进行检测,针对缺陷信号特点,提出基于分形理论的摩擦焊超声检测缺陷信号的识别方法.运用一种新的分形维数计算方法一归一化尺度计盒维数法对试件超声回波缺陷信号盒维数进行计算,统计分析出三种缺陷信号盒维数范围.并对结果进行方差分析.分析结果表明,有缺陷信号与无缺陷信号有着各自分形维数区间,且不交叠,据此可用来判断缺陷的有无.该方法对摩擦焊件缺陷信号的分类识别有较好的统计特性.     

基于特高频法的GIS局部放电典型缺陷检测及模式识别

GIS局部放电检测是GIS绝缘状态监测的一种科学有效的方法。当今,局部放电信号类型的识别越来越受到电气试验技术人员的重视。现基于特高频检测方法,对GIS局部放电中几种典型缺陷的检测及模式识别进行研究。     

基于表面灰度图像的植物磨料磨损表面分形维数计算

以磨损表面SEM灰度图像为研究对象,采用变分法、双毯法、差分盒计数法和盒计数法4种方法计算植物磨料对不同金属材料磨损表面的三维分形维数。结果表明:植物磨料对金属材料磨损表面具有分形特征,分形维数与体积损失有着密切关系,体积磨损越大,分形维数也越大。4种算法中,变分法计算的分形维数稳定性最好,其次是双毯法和差分盒计数法,盒计数法对表面噪声最为敏感。     

GIS发生局部放电的原因分析及在线监测研究

局部放电既是GIS绝缘劣化的征兆和表现形式,又是绝缘进一步劣化的原因。对GIS进行局部放电检测能够有效地发现其内部早期的绝缘缺陷,以便采取措施,避免其进一步发展,提高GIS的可靠性。文章分析了GIS产生放电的原因,介绍了GIS局部放电的在线监测方法。     

基于Wigner分布和分形维数的柴油机故障诊断

针对柴油机配气机构故障诊断问题,提出了一种基于Wigner分布和差分分形盒维数的故障诊断方法。首先,利用改进局部均值分解算法对柴油机缸盖振动信号进行分解,并采用相关性分析剔除噪声和伪分量;然后,分别对各相关分量进行Wigner时频分析,将结果线性叠加得到振动时频图,再提取图像的差分分形盒维数作为故障特征;最后,利用k-最近邻(k-NN)实现故障诊断。仿真结果表明,改进局部均值分解算法可以抑制Wigner分布交叉项的干扰。实验结果显示,差分分形盒维数优于其他6种典型故障特征,利用本研究提出的方法对配气机构进行故障诊断的正确率为97.2%,该方法可以用于柴油机配气机构故障诊断。     

非金属夹杂物分形特征研究

材料的宏观力学性能与材料的微观组织密切相关，以显微组织中离散的点状、线状分布的典型夹杂物为研究对象，用计盒维数测量了标准图谱中不同等级的非金属夹杂物图形的分形维数，考察了夹杂物等级与分维之间的关系。结果表明，非金属夹杂物可用分形维数定量描述；随着夹杂物等级的上升，分形维数变大。     

基于加权SVM的高压GIS设备绝缘缺陷检测方法

常规的高压GIS设备绝缘缺陷的检测方法中,通常使用红外热线的方法来检测缺陷,导致检测精度较低,提出基于加权SVM的高压GIS设备绝缘缺陷检测方法。在方法的整体设计流程中,首先通过局部放电信号确定GIS设备缺陷特征,然后在相位周期内统计特征参数并对特征进行降维,进而基于加权SVM建立检测模型,通过模型运算取得最优解,实现设备绝缘缺陷的检测。实验结果表明所设计的检测方法具有较高的精度,可以运用到实际的缺陷检测流程中。     

裂纹圆盘的动态分形特征及裂纹诊断识别

运用有限元方法对裂纹圆盘进行了动态分析，获得了其振型和位移响应；应用多重分形理论计算和分析了裂纹圆盘的振型、位移响应的广义维数和敏感维数，把广义分形维数、敏感维数作为裂纹圆盘故障的特征量，对裂纹圆盘的裂纹诊断与识别进行了定性及定量研究；针对裂纹圆盘的振型、位移响应的广义维数，提出在局部范围内，用局部线性法对圆盘裂纹长度进行定量识别，得到了定量识别裂纹长度的计算公式，通过实例计算考查了裂纹长度的计算精度。分析计算表明，用局部线性法对裂纹长度进行定量识别具有一定的实用性。     

PD pattern recognition based on multi-fractal dimension in GIS

This paper designs four types of gas insulated substation (GIS) defect models based on partial discharge (PD) characteristics and its defections. The GIS gray intensity images are constructed based on the mass specimens gathered by the ultra-high frequency and high-speed sampling systems. The multi-fractal dimension is founded on the box-counting dimension and multi-fractal theories. The GIS gray intensity images distillation methods, based on multi-fractal characteristics, is put forward. The box-counting dimension, multi-fractal dimension, and discharge centrobaric characteristics of the PD images are also extracted. The characteristic variables are then classified by the radial basis function (RBF) network. Identified results show that the methods can effectively elevate the discrimination of the four types of defects in GIS.     

PD pattern recognition based on multi-fractal dimension in GIS

This paper designs four types of gas insulated substation (GIS) defect models based on partial discharge (PD) characteristics and its defections. TheGIS gray intensity images are constructed based on the mass specimens gathered by the ultra-high frequency and high-speed sampling systems. The multi-fractal dimension is founded on the box-counting dimension and multi-fractal theories. The GIS gray intensity images distillation methods, based on multi-fractal characteristics, is put forward. The box-counting dimension, multi-fractal dimension, and discharge centrobaric characteristics of the PD images are also extracted. The characteristic variables are then classified by the radial basis function (RBF) network. Identified results show that the methods can effectively elevate the discrimination of the four types of defects in GIS. __________ Translated from Chinese Journal of Scientific Instrument, 2007, 28(4): 597–601 [译自: 仪器仪表学报]     

基于小波包和分形盒维数的滚动轴承故障诊断

为诊断滚动轴承不同部件产生的故障,针对轴承故障信号具有非线性、非平稳振动的特点,运用小波包和分形理论,定量计算了滚动轴承不同部件故障信号及小波包重构信号的盒维数。实验结果表明,滚动轴承不同的故障类型具有不同的盒维数。正常滚动轴承盒维数最大,依次为滚珠故障盒维数、内环故障盒维数,外环故障盒维数最小。分形盒维数能定量地识别滚动轴承不同部件的故障,提高滚动轴承故障诊断的准确率,为滚动轴承智能故障诊断提供可靠依据。     

A fractal-dimension-based signal-processing technique and its use for nondestructive testing

A fractal-dimension-based signal-processing technique has been extensively applied to various fields, but the use of the method to characterize discrete time-domain ultrasonic signals reflecting defects and any other structural-material inhomogeneities has not been fully investigated. The fractal features of the ultrasonic echoes with fractal dimensions and their implementation in nondestructive testing are investigated. In order to obtain a faithful representation of the fractal dimensions, two improved fractal dimension algorithms are presented: the box-counting method and the R/S (range/standard deviation) method. Their capabilities are evaluated with two kinds of fractal signals: the FBM (fractal Brownian motion) and WM (Weierstrass-Mandelbrot) signals. A new method to guarantee the feasibility of the calculated fractal dimensions is proposed on the basis of the analysis of the results simulated above. Then, the fractal dimensions of ultrasonic signals measured from a pipeline sample and from carbon-steel and aluminum specimens are calculated and statistically analyzed to find the fractal properties of the ultrasonic signals. The experimental results show that ultrasonic signals have the property of scale invariance that the fractal set possessed. The fractal dimension is indicative of the complexity and degree of irregularity of the waveform of an ultrasonic signal. The fractal dimensions of ultrasonic signals from various defects and microstructures are found to possess solid distribution intervals, which can be used to identify the presence of defects and the features of materials. The potential of the technique for testing defects and assessing the microstructure of materials via the use of ultrasonic echoes is revealed. The text was submitted by the authors in English.     

苜蓿草粉对金属材料磨损表面的分形维数计算

基于分形理论和扫描电镜分析,采用差分计数盒法(DBC)计算了苜蓿草粉对金属材料磨损表面的三维分形维数。结果表明:基于DBC法和磨损表面SEM照片计算所得的分形维数,可以表征苜蓿草粉对金属材料磨损表面的形貌特征;相同加工条件下的金属加工表面具有相近的分形维数,但不能通过分形维数区分材料;苜蓿草粉磨损后的磨损表面分形维数与磨损体积损失有着密切的关系,体积磨损量越大,分形维数也越大。     

铁谱磨粒多重分形特征研究

为有效描述铁谱磨粒特征,提出用多重分形谱参数表达磨粒形态特征的新方法。选择盒计数法计算磨粒图像的多重分形谱,研究磨粒多重分形谱的有效性,分析磨粒多重分形谱参数的不变性和鲁棒性;确定磨粒图像预处理方法,并对4类典型磨粒的多重分形谱参数进行统计分析。结果表明:将多重分形谱参数应用于磨粒识别,总识别率为82.5%。磨粒具有明显的多重分形特性,可用多重分形谱参数来描述磨粒的形态特征;多重分形谱参数具有平移不变性,但对灰度变化和噪声干扰的鲁棒性较差,在提取多重分形谱参数时,需要对磨粒图像做严格的预处理。     

基于动应力分形特征的薄壁圆盘裂纹诊断识别研究

运用有限元方法对裂纹圆盘进行了动态特性分析,获得了振型和应力响应。应用多重分形理论计算和分析了裂纹圆盘的振型、应力响应的广义维数和敏感维数,把广义分形维数、敏感维数作为裂纹圆盘故障的特征量,对裂纹圆盘的裂纹诊断与识别进行了定性及定量研究。通过计算裂纹圆盘的振型、应力响应的广义维数相关系数,用广义维数相关系数法对圆盘裂纹长度进行定量识别。分析计算表明,广义维数相关系数法对裂纹长度的定量识别具有一定的实用性。     

Pest Detection Method Using Multi-Fractal Analysis

A novel method for pest detection is proposed based on the theory of multi-fractal spectrum to extract pests on plant leaves.Both local and global information of the image regularity were obtained by multi-fractal analysis.By applying fractal dimension,the spots on leaf images can be extracted without loosing or introducing any information.The different parts of images are re-analysis by morphology operations to determine the candidate regions of pests.The performance of multi-fractal analysis of whitefly detection is investigated through greenhouse experiments.The experimental results show that the proposed method is robust to noise from light and is not sensitive to the complex environment.