基于功率谱密度分析的脉冲涡流缺陷分类法

在巨磁阻脉冲涡流传感器(GMR-PEC)上实现平板导体表面和次表面裂纹缺陷以及孔缺陷进行精确分类。在频率分析基础上,提出了一种新的缺陷特征量——涡流差分响应信号的功率谱密度。由于主成分分析具有良好的降维特性,采用主成分分析结合线性判别分类(PCALDA)和贝叶斯分类(PCA-Bayes)进行缺陷的分类。结果表明,基于新的特征量的分类方法能实现导体表面和次表面的裂纹和孔缺陷的精确分类,在脉冲涡流自动测量领域具有潜在的意义。     

Ultrasonic NDE discrimination with the gradient descent algorithm and SAFT image processing

Ultrasonic inspection methods are widely used for detecting flaws in materials. One of the most popular methods involves the extraction of an appropriate set of features followed by the use of a neural network for the classification of the signals in the feature space. This paper describes an approach which uses LMS method to determine the coordinates of the ultrasonic probe followed by the use of SAFT to estimate the location of the ultrasonic reflector. The method is employed for discriminating NDE signals from the steam generator tubes in a nuclear power plant. The results obtained by using this scheme, of discriminating the ultrasonic signals from cracks and deposits within steam generator tubes, are presented.     

Assortment of ultrasonic nondestructive evaluation signals using the centroid searching method with the LMS algorithm and SAFT

NDE using ultrasonic signals is a very useful technique for the assessment of solid materials, construction, food, and biomedicine. Among many NDE methods, the ultrasonic inspections may involve the extraction of an appropriate set of features or a neural network for the classification of the signals in the feature space. This paper presents an approach that uses a geometric method and the LMS (Least Mean Square) algorithm to determine the coordinates of the ultrasonic probe followed by the SAFT (Synthetic Aperture Focusing Technique) and centroid searching technique to estimate the location of the ultrasonic reflector. The proposed method is employed to classify ultrasound NDE signals from cracks and deposits within steam generator tubes in a nuclear power plant.     

Study on defect classification in multi-layer structures based on Fisher linear discriminate analysis by using pulsed eddy current technique

Pulsed eddy current (PEC) is an emerging non-destructive testing technique with wide application potential. In this study, defect parameter identification in multi-layer structures is examined by using the PEC technique, and a Fisher linear discriminate analysis (FLDA)-based defect classification method is proposed. Defect localization and shape identification are investigated, and defects on the surface and subsurface of the third layer are discriminated. The time domain characterization method is introduced and researched by using the peak time and zero-crossing time of PEC response signals, the principal component analysis algorithm and the FLDA method. The feature extraction results of the three methods are used as the input values of support vector machine for defect classification and feature extraction, and the classification methods are compared. Theoretical analysis and experimental results show that the proposed method can contribute to effective classification for defect parameter identification.     

基于PSO-PNN的铸坯全长表面纵裂纹在线预测

表面纵裂纹是铸坯质量缺陷中一种最常见的表面质量缺陷。环境因素使得铸坯表面纵裂纹在线检测精度不高,各大钢厂铸坯质检仍要依赖人工,因此提出一种基于粒子群PSO优化概率神经网络PNN的铸坯全长表面纵裂纹预测方法。首先,建立铸坯生产过程跟踪及数据时空变换模型,构建铸坯生产系统将生产过程数据与铸坯长度方向进行匹配;再利用PNN的Bayes 最小风险准则进行有监督特征学习,并利用寻优算法PSO优化PNN关键参数的选取,得到最终的模型PSO-PNN;最后,利用某钢厂连铸产线铸坯质量缺陷数据和生产过程数据进行试验验证。结果表明,该方法对铸坯整体的质量预测分类精度达到97.5%,铸坯全长的裂纹缺陷的预测精确率和召回率均在92%以上,能有效实现铸坯全长表面纵裂纹的预测,为现场质检人员提供参考。     

Classification of pulsed eddy current GMR data on aircraft structures

This paper presents a technique to automatically detect third-layer cracks at rivet sites in aircraft structures using the response signals collected by giant magneto-resistive (GMR) sensors. The inspection system uses pulsed waveform as the excitation source of a multi-line coil and captures the transient fields associated with the induced eddy currents via a GMR sensor, which was developed to detect cracking and corrosion in multi-layer aircraft structures. An automatic scan of the region around the rivet generates C-scan image data that can be processed to detect cracks under the rivet head. Using a 2-D image of each rivet head, feature extraction and classification schemes based on principal component analysis and the k-means algorithm have been successfully developed to detect cracks of varying size located in the third layers at a depth of up to 10 mm below the surface.     

石化齿轮箱振动信号高价值量纲一化特征提取方法研究

石化齿轮箱故障信号特征之间呈现模糊性和耦合性,正常状态与异常状态之间相互影响,导致故障分类难。为解决此问题,提出一种石化齿轮箱振动信号高价值量纲一化特征提取方法。运用数据预处理方法,从采集的数据中抽取有效数据,对其进行量纲一化指标表示,再依据量纲一化指标分布规律提取高价值量纲一化特征。为保证样本的完备性,建立高价值量纲一化特征的增容模型。利用该方法能够准确提取石化齿轮箱实验机组轴承故障特征,可为石化齿轮箱特征提取提供参考。     

基于振动时域特征的船用滚动轴承故障诊断方法

基于机器学习故障诊断方法,针对船用滚动轴承复合故障特征提取多样化的特点,提出一种以振动信号时域指标为特征的随机森林故障诊断方法。将振动时域信号进行清洗转换,构造5个量纲一化指标的衍生特征,并选取以决策树为基本分类器的随机森林算法建立训练模型;通过特征筛选、评估测试和模型优化得到较为理想的故障诊断分类模型;采用滚动轴承竞赛数据集进行模型仿真,并结合实际模拟8种船用滚动轴承故障状态。通过三向振动实验和算法建模,证明特征提取的科学性和故障诊断模型的有效性。结果表明:采用该方法,数据仿真诊断准确率为98.61%,实验诊断准确率为98.85%,且该方法在振动采集方向为轴向时诊断效果最优。     

Improving feature ranking for biomarker discovery in proteomics mass spectrometry data using genetic programming

Feature selection on mass spectrometry (MS) data is essential for improving classification performance and biomarker discovery. The number of MS samples is typically very small compared with the high dimensionality of the samples, which makes the problem of biomarker discovery very hard. In this paper, we propose the use of genetic programming for biomarker detection and classification of MS data. The proposed approach is composed of two phases: in the first phase, feature selection and ranking are performed. In the second phase, classification is performed. The results show that the proposed method can achieve better classification performance and biomarker detection rate than the information gain- (IG) based and the RELIEF feature selection methods. Meanwhile, four classifiers, Naive Bayes, J48 decision tree, random forest and support vector machines, are also used to further test the performance of the top ranked features. The results show that the four classifiers using the top ranked features from the proposed method achieve better performance than the IG and the RELIEF methods. Furthermore, GP also outperforms a genetic algorithm approach on most of the used data sets.     

Greedy information acquisition algorithm: A new information theoretic approach to dynamic information acquisition in neural networks

In this paper, we proopose a new information theoretic approach to competitive learning. The new approach is called greedy information acquisition , because networks try to absorb as much information as possible in every stage of learning. In the first phase, with minimum network architecture for realizing competition, information is maximized. In the second phase, a new unit is added, and thereby information is again increased as much as possible. This proceess continues until no more increase in information is possible. Through greedy information maximization, different sets of important features in input patterns can be cumulatively discovered in successive stages. We applied our approach to three problems: a dipole problem; a language classification problem; and a phonological feature detection problem. Experimental results confirmed that information maximization can be repeatedly applied and that different features in input patterns are gradually discovered. We also compared our method with conventional competitive learning and multivariate analysis. The experimental results confirmed that our new method can detect salient features in input patterns more clearly than the other methods.     

Diagnosis of tapping process using spindle motor current

A two-stage pair-wise feature selection and classification method is proposed to diagnose the tapping process using the electrical current signal of the spindle motor. The motor current is measured with a non-intrusive Hall effect sensor. The diagnostic process is divided into two stages. In the first stage, wavelet coefficients of the current signals are computed and a subset of these coefficients is chosen as the initial diagnostic features based on their discriminatory powers for each pair of the fault classes. At the second stage, principal component analysis (PCA) is applied to reduce the dimensionality of the initial diagnostic features. Non-linear radial basis probabilistic neural networks (RPNNs) are employed for pair-wise feature classification. The final diagnostic decision is made based on a statistical voting procedure. The proposed method has been demonstrated with experimental data. On average a 93% success rate has been achieved.     

Investigation on eddy current pulsed thermography to detect hidden cracks on corroded metal surface

This paper explores the feasibility of eddy current pulsed thermography (ECPT) to detect hidden cracks on corroded metal surface without removing the corrosion layer. The detection mechanisms are analyzed by using electromagnetic and heat conduction theories. Experiments are conducted on a metallic bar with three hidden cracks and the validity of ECPT is verified with the analysis of IR images and thermal responses. In order to further improve the detection sensitivity of ECPT, the principal component analysis (PCA) is applied to enhance the features of hidden cracks in the raw IR images by eliminating the effects of uneven corrosion and non-uniform heating. It is found that ECPT combined with PCA provides a convenient and effective way to detect hidden cracks on corroded metal surface.     

涡流检测的信号处理技术

介绍涡流检测信号处理技术中特征值的提取方法，以及利用神经网络和信息融合技术进一步对信号进行分类、识别和评价的技术发展，还介绍了涡流逆问题求解的发展趋势和根据信号特点选择信号的处理方法。     

基于Contourlet和KPCA的焊接缺陷图像特征提取

为了进一步提高焊接缺陷识别的准确度和效率,提出了一种基于Contourlet变换和混沌粒子群优化核主成分分析(kernel principal component analysis,KPCA)的焊接缺陷图像特征提取方法.首先通过Contourlet变换将焊接缺陷图像进行多尺度分解,提取低频分量和特定方向上的高频分量;然后运用混沌粒子群优化后的KPCA分别提取缺陷训练样本和缺陷测试样本的特征;最后根据测试样本特征与训练样本特征之间的欧式距离确定缺陷测试样本的类型.结果表明,与基于核主成分分析特征提取法、基于小波的核主成分分析特征提取法相比,文中方法提取的特征更为完整,识别率更高,运行速度较快.     

Feature extraction and selection for defect classification of pulsed eddy current NDT

Pulsed eddy current (PEC) is a new emerging nondestructive testing (NDT) technique using a broadband pulse excitation with rich frequency information and has wide application potentials. This technique mainly uses feature points and response signal shapes for defect detection and characterization, including peak point, frequency analysis, and statistical methods such as principal component analysis (PCA). This paper introduces the application of Hilbert transform to extract a new descending feature point and use the point as a cutoff point of sampling data for detection and feature estimation. The response signal is then divided by the conventional rising, peak, and the new descending points. Some shape features of the rising part and descending part are extracted. The characters of shape features are also discussed and compared. Various feature selection and integrations are proposed for defect classification. Experimental studies, including blind tests, show the validation of the new features and combination of selected features in defect classification. The robustness of the features and further work are also discussed.     

Matrix factorization methods: Application to thermal NDT/E

A typical problem in thermal nondestructive testing/evaluation (TNDT/E) is that of unsupervised feature extraction from the experimental data. Matrix factorization methods (MFMs) are mathematical techniques well suited for this task. In this paper we present the application of three MFMs: principal component analysis (PCA), non-negative matrix factorization (NMF), and archetypal analysis (AA). To better understand the peculiarities of each method the results are first compared on simulated data. It will be shown that the shape of the data set strongly affects the performance. A good understanding of the actual shape of the thermal NDT data is required to properly choose the most suitable MFM, as it is shown in the application to experimental data.     

基于深度神经网络的粗糙度分类检测

现有的机器视觉测量表面粗糙度的方法依赖于特征提取来量化表面形态并建立预测模型,并且图片样本采集时高度依赖光源环境和拍摄角度。提出利用Swin-Transformer模型直接从表面纹理的数字图像中评估表面粗糙度并进行分类,避免了特征提取。实验中采用了不同的光源亮度和不同的角度进行拍摄,旨在模拟工业生产在线检测的环境。实验分类结果表明该方法对光源和拍摄角度具有良好的鲁棒性,测试集样本准确率可达97.54%。     

焊接机器人轨迹跟踪研究现状

各类传感器和智能控制方法极大促进了机器人在焊缝跟踪中的应用,不仅提高了焊缝跟踪的精度,同时提高了焊接效率和保证了焊接质量。简述了机器人焊缝跟踪系统的结构,详述了焊缝跟踪过程中各类传感器的工作原理及其特点;阐述了图像处理技术在机器人焊缝轨迹跟踪过程中的研究进展,并对图像的预处理、图像分割与边缘检测和特征提取等研究方法进行了分析。最后,总结了智能控制方法在焊缝跟踪中研究进展及不同形状的焊缝跟踪情况。     

固体火箭发动机高能X射线探伤胶片影像三维重建

固体火箭发动机是导弹武器系统的主要动力装置,需要定期进行无损检测。常规的检测方法是对X射线照相得到的二维胶片进行评判,然而利用二维图像来反映三维空间结构,在直观显示上存在很大的缺陷性。介绍了利用二维胶片图像还原发动机的空间三维结构的基本重建原理和六大体系结构,即数字图像的获取、各项参数的标定、图像的预处理和特征提取、图像的立体匹配以及图像信息三维重建。