Learning material defect patterns by separating mixtures of independent component analyzers from NDT sonic signals

This paper introduces the application of independent component analysis mixture modelling (ICAMM) in non-destructive testing (NDT). The application consists of discriminating patterns for material quality control from homogeneous and defective materials inspected by impact-echo testing. This problem is modelled as a mixture of independent component analysis (ICA) models, representing a class of defective or homogeneous material by an ICA model whose parameters are learned from the impact-echo signal spectrum. These parameters define a kind of particular signature for the different defects. The proposed procedure is intended to exploit to the maximum the information obtained with the cost efficiency of only a single impact. To illustrate this capability, four levels of classification detail (material condition, kind of defect, defect orientation, and defect dimension) are defined, with the lowest level of detail having up to 12 classes. The results from several 3D finite element models and lab specimens of an aluminium alloy that contain defects of different shapes and sizes in different locations are included. The performance of the classification by ICA mixtures is compared with linear discriminant analysis (LDA) and with multi-layer perceptron (MLP) classification. We demonstrate that the mass spectra from impact-echo testing fit ICAMM, and we also show the feasibility of ICAMM to contribute in NDT applications.     

A novel and effective multistage classification system for microscopic starch grain images

This article presents a novel and effective multistage system for classifying Chinese Materia Medica microscopic starch grain images. The proposed classification system is constructed based on the Gaussian mixture model‐based clustering, the feature assignment algorithm, and the similarity measurement. Several features for each starch grain image are extracted and every class of drug is represented by a set of characteristic features. For each stage of the system, only one feature is chosen and assigned to that stage via the feature assignment algorithm, and the corresponding characteristic features are subdivided into smaller subsets based on clustering techniques. At the final stage, each subset contains a certain class of drugs (with corresponding characteristic features) and similarity measurement is carried out for starch grain classification. Three sets of the current state‐of‐the‐art starch grain features including the granulometric size distribution, the chord length distribution, and the wavelet signature are used to construct the system. Experimental results on a database of 240 images of 24 classes of drugs reveal the superior performance of the multistage system. Comparison with the traditional starch grain classification approaches indicates that our proposed multistage method produces a marked improvement in classification performance. Microsc. Res. Tech. 2009. © 2009 Wiley‐Liss, Inc.     

Supervised locally tangent space alignment for machine fault diagnosis

How to deal with the high-dimensional and nonlinear data is a challenging problem for fault diagnosis. An unsupervised locally tangent space alignment (LTSA) has recently proven to be an effective unsupervised manifold learning algorithm for high-dimensional data analysis. In this paper, a supervised expansion of LTSA (named S-LTSA) is proposed, which takes full advantage of class label information to improve classification performance. Based on S-LTSA, a novel machine fault diagnosis approach is proposed to deal with the high-dimensional fault data that contain multiple manifolds corresponding to fault classes. The experiment results with bearing fault data show that the proposed approach outperforms the other fault pattern recognition approaches such PCA, ICA, LDA and LTSA.     

整体改进的基于支持向量机的故障诊断方法

为了消除噪声或野值样本对支持向量机分类器推广性能的不利影响,从数据预处理、特征提取和分类器设计等几个方面对现有的基于支持向量机的故障诊断方法进行了整体改进。一方面,在独立分量分析的基础上提出一种残余总体相关分析时域特征提取方法,利用独立分量分析的冗余取消特性以及残余总体相关分析的整体约简能力,抽取描述不同故障模式类的典型低维特征,削减原始数据中的噪声干扰;另一方面,对各模式类特征样本进行模糊C-均值聚类,然后以类内平均距离和类间平均距离共同构建一个有效性判别准则,用于区分特征空间中的有效样本与野值点,去除野值对支持向量机目标函数的影响。在此基础上引入具有可控稀化解的前向最小平方近似支持向量机算法,并采用基于复杂多故障模式分级识别的二分类策略,共同形成一种整体改进的基于支持向量机的故障诊断方法。对齿轮箱故障的诊断结果验证了该方法的有效性,对于受强噪声干扰的小样本数据,所构建的故障分类器也具有良好的推广能力。     

基于总体局域均值分解及稀疏表示分类的天然气管道泄漏孔径识别

针对天然气管道泄漏受孔径、传感器距离、管道内压力等多种因素影响,特征提取及识别算法较为复杂的问题,提出了基于总体局域均值分解-相对熵的特征提取算法并结合稀疏表示分类的泄漏孔径识别新方法。该方法采用总体局域均值分解方法对泄漏信号进行自适应分解,得到不同孔径泄漏信号的特征信息,并根据KL散度选择包含主要泄漏信息的PF分量,在此基础上提取多种时频特征参数,获取全面准确表征泄漏信号的特征向量;针对小样本复杂信号的分类,提出稀疏表示分类器实现泄漏孔径准确分类。该分类器采用过完备字典求得测试信号的最稀疏解,并以此解作为测试信号的稀疏重构系数,以获取测试信号在不同类别中的重构信号,最终通过判断测试信号与重构信号的残差值大小完成泄漏孔径分类。实验结果表明,所提出的算法比传统的SVM及BP分类算法识别准确率高。     

谷物干燥机烘干温度对种子发芽率的影响分析

针对谷物干燥机烘干后的种子存在着发芽率降低的问题,本研究对影响种子发芽率的诸多因素进行了归纳,对温度参数进行检测,试验前后的种子发芽率,温度参数对烘干种子发芽率的影响进行了研究,提出了一种谷物干燥机烘干温度分析方法,利用合理布置的比对温度传感器,对谷物干燥机内的实时温度进行了监测。研究结果表明,烘干后种子发芽率降低的主要原因是谷物干燥机的机配温度传感器存在误差,影响烘干温度的准确性;谷物干燥机热风道结构的设计,温度最高位置会导致相邻谷物温度超差;谷物干燥机内温度分布不均匀导致局部谷物温度超差及机配温度传感器在谷物干燥机内的布置不合理,未反应机内最高温度。     

Identification of novel nonedible oil seeds via scanning electron microscopy for biodiesel production

Exploration of substitute energy feed‐stocks is the much‐debated topic in the scientific society due to increasing power crises and related ecological concerns. As a source of sustainable energy, biodiesel turns out to be the best alternative to petro fuels. In this context, nonedible oil‐producing seeds might be a potential source for biodiesel production owing to their environment‐friendly nature and cost‐effectiveness. The current study, consequently, deals with the investigation and identification of micro‐morphological characters between six novel nonedible oil‐bearing seeds employing scanning electron microscopy as possible biodiesel feed‐stocks. Light microscopic examinations show that seed size varies from 0.3 to 1.3 cm in width and 0.5 to 1.5 cm in. Additionally, a large difference in seed color ranges from dark brown, black, and various shades of light brown was also witnessed. The FFA content of the seeds ranges in 0.3–4.1 mg KOH/g, and the seed oil content fall in 30–65% (w/w) range. SEM‐mediated seed ultrastructure investigations displays greater variation in seed size, shape, color, periclinal wall shape, and sculpturing and so on. All the seeds differ from rounded, ovoid, ovate, oblong, flattened, and elliptical shape. Greater variation in seed wall structure has been seen from angular, entire, irregular, straight, elongated, smooth, and polygonal. The periclinal wall arrangements show alteration from flat, depressed, elevated, smooth, pentagonal, bullate, and coarse seed margins. The results obtained from the current study suggest that scanning electron microscopy could be a beneficial tool in vitalizing the hidden micromorphological characters among various nonedible oil producing seeds, which eventually helps in exploration, correct identification, seed classification, and authentication in future.     

A new approach to detection of defects in rolling element bearings based on statistical pattern recognition

The paper presents a new approach to the classification of rolling element bearing faults by implementing statistical pattern recognition. Diagnostics of rolling element bearing faults actually represents the problem of pattern classification and recognition, where the key step is feature extraction from the vibration signal. Characterization of each recorded vibration signal is performed by a combination of signal's time-varying statistical parameters and characteristic rolling element bearing fault frequency components obtained through the envelope analysis method. In this way, an 18-dimensional vector of the vibration signal feature is obtained. Dimension reduction of the 18-dimensional feature vectors was performed afterward into two-dimensional vectors representing the training set for the design of parameter classifiers. The classification was performed in two classes, into defective and functional rolling element bearings. Main trait of parameter classifiers is simplicity in their design process, as opposed to classifiers based on neural networks, which employ complex training algorithms.     

A frequency response based structural damage localization method using independent component analysis

Vibration-based structural damage detection has been the focus of attention by many researchers over the last few decades. However, most methods proposed for this purpose utilize extracted modal parameters or some indices constructed on the basis these parameters. Our literature review revealed that few papers have employed frequency response functions (FRFs) for detecting structural damage. In this paper, a technique is presented for damage detection which is based on measured FRFs. Independent component analysis (ICA) has been implemented on the spatiotemporal responses in each frequency in order to reduce the dimension of the data. This is based on the concept that the forced harmonic response of a linear vibrating system can be fully captured utilizing a single ICA mode. A different approach is also presented in which ICA is applied to the frequency domain data. Operational deflection shapes (ODSs) have been decomposed using ICA to localize the damage. The efficiency of both methods is demonstrated through some numerical and experimental case studies.     

立式浅盆型复合种盘大豆排种器结构设计与参数优化

为满足机械式精密排种器高速作业的要求,以增加充填力的方式对立式圆盘排种器进行改进,设计了一种利用重力与离心力以及种子间相互作用力作为复合力进行充种的双腔立式浅盆型复合种盘机械式精密排种器,并对排种盘的特征参数进行了优化。利用离散元仿真分析软件EDEM对立式浅盆型种盘进行了以排种盘直径、折边倾角、作业速度为影响因素,充填率为性能指标的三因素五水平二次正交旋转组合仿真设计,并运用Design-expert 8.0软件对数据进行了处理,获得了排种盘最优的结构参数组合,同时利用物理样机试验进行了验证。研究结果表明：立式浅盆型种盘的最佳结构参数组合为排种盘直径234mm、折边倾角68°,在8 ~12km/h高速作业时,合格率均能达到90%以上。研究结果为产品定型提供了定量的依据。     

Combined Cellular Automaton Model for Dynamic Recrystallization Evolution of 42CrMo Cast Steel

The dynamic recrystallization(DRX) simulation performance largely depends on simulated grain topological struc?tures. However, currently solutions used di erent models for describing two?dimensional(2 D) and three?dimensional(3 D) grain size distributions. Therefore, it is necessary to develop a more universal simulation technique. A cellular automaton(CA) model combined with an optimized topology deformation technology is proposed to simulate the microstructural evolution of 42 CrMo cast steel during DRX. In order to obtain values of material constants adopted in the CA model, hot deformation characteristics of 42 CrMo cast steel are investigated by hot compression metal?lographic testing. The proposed CA model deviates in two important aspects from the regular CA model. First, an optimized grain topology deformation technology is utilized for studying the hot compression e ect on the topology of grain deformation. Second, the overlapping grain topological structures are optimized by using an independent component analysis method, and the influence of various thermomechanical parameters on the nucleation process, grain growth kinetics, and mean grain sizes observed during DRX are explored. Experimental study shows that the average relative root mean square error(RRMSE) of the mean grain diameter obtained by the regular CA model is equal to 0.173, while the magnitude calculated using the proposed optimized CA model is only 0.11. This paper pro?poses a novel combined CA model for simulating the microstructural evolution of 42 CrMo cast steel, which notably uses a ICA?based grain topology deformation method to optimize the overlapping grain topological structures in simulation.     

基于机器视觉的汽车换挡面板自动分类方法

针对报废汽车企业对不同型号汽车换挡面板自动分类的需求,提出了一种基于机器视觉的面板自动分类方法。该方法首先使用Blob分析对图像进行预处理,分割出图像中的换挡面板目标对象,接着提取被分割图像的11个形状特征作为特征向量,利用特征向量训练多层感知器(Multi-Layer Perceptron,MLP),最后使用完成训练的MLP神经网络模型对不同类型的汽车换挡面板进行分类预测。实验结果表明,该方法的分类准确率达到99%,每幅图像的分类时间小于100 ms,验证了该方法的有效性与高效性,能够作为报废汽车换挡面板自动分类的一种手段。     

基于局部均值分解与拉普拉斯特征映射的滚动轴承故障诊断方法

针对滚动轴承非平稳振动信号的特征提取及维数优化问题,提出了融合局部均值分解与拉普拉斯特征映射的轴承故障诊断方法。首先,通过局部均值分解对非平稳振动信号进行平稳化分解,提取乘积函数分量、瞬时频率及瞬时幅值的高维信号特征集;然后,将高维特征集作为拉普拉斯特征映射算法的学习对象,提取轴承高维故障特征集的内在流形分布,以获得敏感、稳定的轴承振动特征参数,实现基于非平稳振动信号分析的滚动轴承故障特征提取;最后,结合支持向量分类模型量化LMD-LE方法的特征提取效果,实现不同状况下的轴承故障分类。轴承故障样本分类识别平均正确率达到91.17%,表明LMD-LE方法有效实现了高维局部均值分解特征集合的降噪,所提取的特征矩阵对轴承故障特征描述准确。     

Shape and texture features in the automated classification of adhesive and abrasive wear particles

In this study the automated classification system, developed previously by the authors, was used to classify wear particles. Two kinds of wear particles, adhesive and abrasive, were classified. The wear particles were generated using a pin-on-disk tribometer. Various operating conditions of load, sliding time and abrasive grit size were applied to simulate adhesive and abrasive wear of different severity. SEM images of wear particles were acquired, forming a database for further analysis. The particle images were divided into eight groups or classes, each class representing different wear test conditions. All eight particle classes were first examined visually. Next, area, perimeter and elongation parameters were determined for each class and the parameters were statistically analysed. The automated classification system, based on particle surface texture, was then applied to all particle classes. The results of the automated particle classification were compared to those based on either the visual assessment of particle morphology or numerical parameter values. It was shown that the texture-based classification system was a more efficient and accurate way of distinguishing between various wear particles than classification based on size and shape of wear particles. It seems that the texture-based classification method developed has great potential to become a very useful tool in the machine condition monitoring industry.     

Optimising the thermoforming process of polymeric foams: an approach by using the Taguchi method and the utility concept

The use of the conventional Taguchi method for determining the optimum setting of controllable factors through off-line experiments focuses on products with a single quality characteristic or response. However, most products have several qualitative characteristics or responses of interest. The Taguchi method in itself optimises a single response or performance characteristic, yielding a set of process parameters. This particular setting, however, may not give the desired results for other characteristics of the product. There is a need to obtain a single optimum setting of process parameters that can be used to produce products with optimum or near optimum quality characteristics as a whole. Multi-characteristic response optimisation may be the solution to the above problem. In this report, a case study on thermoforming polypropylene foams, utilising a simplified multi-criterion methodology based on Taguchis approach and utility concept, is discussed. Key processing factors affecting product quality are identified. It has been shown that the proposed Taguchi approach with the utility concept can provide an appropriate solution to yield a satisfactory product quality for a multi-response process optimisation problem.     

Microscopic techniques for characterization and authentication of oil-yielding seeds

Investigation of alternative energy sources is need of current time due to growing power crisis and associated environmental issues. Biodiesel is considered as sustainable power source and promising alternative to fossil fuels. Therefore, our current investigation aimed to identify micromorphological characters of 10 novel nonedible oil-yielding seeds through scanning electron microscopy. It was revealed from light microscopic study that there is variation in seed size from 3 to 15 mm in length and 2 to 11 mm in width. Likewise, a huge variation in color was observed such as light green, greenish yellow, blackish brown, and various shades of brown. Presence and absence of Hilum was observed, and compression of seeds varied from depressed, lateral, and dorsoventral. Seed's shape differs from ovate, clavate, triangular ovate, cuneiform, ovoid, and elliptical shape. Seed oil content fall in range of 18–58% (wt/wt). Free fatty acid content of the seeds varies from 0.3 to 3.1 mg KOH/g. Ultrastructure of seeds exhibited huge variation in shape, size, periclinal wall, anticlinal wall, and surface ornamentation. Nonedible seeds varied in wall structure from angular, wavy, dentate entire, irregular, puzzled, elongated, even, and polygonal. The periclinal wall arrangements show alteration from flat, looped, raised, depressed, lofty, even, pentagonal, polygonal, and undulate seed margins. Outcomes of this investigation recommended that scanning electron microscopy could act as a helpful tool in disclosing the hidden micromorphological characters among nonedible oil-yielding seeds and subsequently helping in correct, authentic seed identification and classification as potential feedstock for biodiesel.     

真空吸附式穴盘精量播种机的研制

穴盘精密播种是现代育苗技术的一重要环节,由于蔬菜、花卉等作物种子具有体积小、重量轻、形状又不规则的特点,给其精密播种带来了困难。为了较好地实现这类种子的精密播种,我们提出并设计了一种真空吸附精量播种机。通过对小白菜、西红柿、黄瓜等作物种子的初步试验,其单粒精播率达90%,漏播率低于5%,说明该机具有较高的播种精度和对不同类型种子良好的适应性,并且整机结构简单,自动化程度高。     

Macromorphological and micromorphological seed features of selected taxa of Caesalpiniaceae using light and scanning electron microscopy

Seed micromorphological and macromorphological characteristics of 12 taxa of Caesalpiniaceae from Pakistan have been studied, using light and scanning electron microscopy, to investigate the importance of seed coat features as a taxonomic tool. Great variations have been observed in color, shape, dimension, and seed surface pattern among the different genera of the family. A taxonomic key was prepared for the studied taxa. Six types of seed shapes were observed; circular, elliptical, irregular, oblong, oval, and ovoid. All examined seeds were hard except fragile seeds of Haematoxylom campechianum. Eight types of surface ornamentation have been noticed that include levigate, lugose, papillate, reticulate, reticulate irregular, reticulate regular, rhombus, and rogues. Majority of the taxa has been observed with thick ornamentation wall but thin ornamentation wall has also been recorded in few species. Fracture line of the various patterns was present in all taxa except genus Bauhinia. Three types of texture crudeness; coarse, medium, and fine have been recorded. Both micromorphological and macromorphological characters of seed are very fruitful in identification and classification of Caesalpiniaceae.     

Identification of novel,non-edible oil seeds via scanning electron microscopy as potential feedstock for green synthesis of biodiesel

In the present era, environmental glitches associated with extensive emission of greenhouse gases (GHG) and energy crises caused by exhausting fossil fuel reservoirs has diverted researcher's interest toward alternative and renewable energy sources. Biodiesel is a renewable, biodegradable, and sustainable alternative to petro-diesel. Biodiesel synthesized from non-edible seed oils is preferred due its cost effectiveness and eco-friendly nature. Hence, our present study focused on investigation and identification of micromorphological characters of six novel, non-edible seed oil feedstock for biodiesel production via scanning electron microscopy (SEM). Results of light microscopy of seeds revealed distinct variation in seed size (15.8–1.8 mm in length and 9.4–1.1 in width), shape (round to Cuneiform), and color (from black to yellowish green). Non-edible seed oil content fall in range of 28–38% (wt/wt). Free fatty acid (FFA) content varied from 0.56 to 2.06 mg KOH/g. Multivariate analysis was performed to investigate correlation between three significant variables of seed oil yielding feedstock such as potential for biodiesel synthesis, oil content, and FFA content via principal component analysis. Ultra morphological investigation of seeds surfaces via SEM exhibited distinctive variation in surface sculpturing, cell arrangement, cell shape, periclinal wall shape, margins, protuberances, and anticlinal wall shape. Seed surface sculpturing varied from reticulate, semitectate, wrinkled, rugose, papillate, perforate, and striate. Periclinal wall arrangements confirmed variation from glabrous, raised, depressed, elevated, smooth, pentagonal, entire, and ripple margins. Whereas, anticlinal walls pattern demonstrated variation from angular, smooth, wavy, deep, dentate, entire, irregular, puzzled, elongated, curved, and depressed. Finally, it was concluded from obtained results that SEM could be a possible useful tool in disclosing veiled micromorphological characters of non-edible oil yielding seeds, which provides useful information to researchers for their correct, authentic identification, and classification in modern synthetic system.     

Modeling blast furnace productivity using support vector machines

Productivity of a modern generation blast furnace was modeled with the help of a leading supervised learning tool viz. Support Vector Machines in the form of (1) minimum error, maximum margin classification function in binary setting of productivity classes (low/high) and (2) the class-specific regression functions for real values of productivity based on epsilon sensitive loss function and minimum regulated risk. The SVMs were trained with large number data-points each of which consisted of a setting of 21 critical input parameters of blast furnace, corresponding productivity value observed, and the productivity class (low/high) attributed. During the training session of the SVMs, the vectors of critical input parameters were required to be mapped into high-dimensional feature space via Radial basis kernel as function and the optimum SVM-RBF classifying function with chosen setting of its hyperparameters that had good generalization property was found using quadratic optimization. The SVM-RBF classifying function could be used to predict the class of productivity (low/high) for any given setting of the critical input parameters. Class-specific SVM-RBF regression models were also developed for both low as well as high-productivity classes and these models could be used to predict real value of productivity for any given setting of the critical input parameters. The SVM-RBF regression model fitted to the high-productivity class was subjected to constrained nonlinear optimization treatment to find the optimum setting of the critical input parameters that gave maximum productivity. The optimum setting of the critical parameters could be used as the target setting obtaining high productivity in the blast furnace.