Machine�Cpart cell formation through visual decipherable clustering of self-organizing map

Machine?Cpart cell formation is used in cellular manufacturing in order to process large varieties, improve quality, and lower work-in-process levels, reducing manufacturing lead time and customer response time while retaining flexibility for new products. This paper presents a new and novel approach for obtaining machine cells and part families. In cellular manufacturing, the fundamental problem is the formation of part families and machine cells. The present paper deals with the self-organizing map (SOM) method, an unsupervised learning algorithm in artificial intelligence which has been used as a visually decipherable clustering tool of machine?Cpart cell formation. The objective of the paper is to cluster the binary machine?Cpart matrix through visually decipherable cluster of SOM color coding and labeling via the SOM map nodes in such a way that the part families are processed in that machine cell. The U-matrix, component plane, principal component projection, scatter plot, and histogram of SOM have been reported in the present work for the successful visualization of the machine?Cpart cell formation. Computational result with the proposed algorithm on a set of group technology problems available in the literature is also presented. The proposed SOM approach produced solutions with a grouping efficacy that is at least as good as any results earlier reported in the literature and improved the grouping efficacy for 70% of the problems and was found to be immensely useful to both industry practitioners and researchers.     

神经网络在非石棉垫片分类中的研究

提出运用神经网络的分类方法来对已知的15种非石棉垫片的配方数据进行分类。分别运用神经网络中的PNN,LVQ和SOM神经网络对其进行分类。结果表明,PNN神经网络和LVQ神经网络在所提供的数据中均能进行有效的分类,而SOM的分类结果则不太理想。     

刀具磨损监测及破损模式的识别

对于金属切削过程中的刀具磨损,提出了基于隐马尔可夫模型的模式识别理论来识别刀具的不同磨损状态,从而预报刀具破损.该方法对切削过程中切削力信号的动态分量和刀柄振动信号进行快速傅里叶变换特征提取,然后利用自组织特征映射对提取的特征矢量进行预分类编码,把矢量编码作为观测序列引入到隐马尔可夫模型中进行机器学习,建立了3个不同磨损状态的隐马尔可夫模型,并利用最大概率进行模式识别.试验表明,该方法对车刀磨损过程进行识别和预报是有效的.     

Applying the self-organization feature map (SOM) algorithm to AE-based tool wear monitoring in micro-cutting

This study applies a self-organization feature map (SOM) neural network to acoustic emission (AE) signal-based tool wear monitoring for a micro-milling process. An experiment was set up to collect the signal during cutting for the system development and performance analysis. The AE signal generated on the workpiece was first transformed to the frequency domain by Fast Fourier transformation (FFT), followed by feature extraction processing using the SOM algorithm. The performance verification in this study adopts a learning vector quantification (LVQ) network to evaluate the effects of the SOM algorithm on the classification performance for tool wear monitoring. To investigate the improvement achieved by the SOM algorithms, this study also investigates cases applying only the LVQ classifier and based on the class mean scatter feature selection (CMSFS) criterion and LVQ. Results show that accurate classification of the tool wear can be obtained by properly selecting features closely related to the tool wear based on the CMSFS and frequency resolution of spectral features. However, the SOM algorithms provide a more reliable methodology of reducing the effect on the system performance contributed by noise or variations in the cutting system.     

Gear fault identification based on Hilbert–Huang transform and SOM neural network

Gear vibration signals always display non-stationary behavior. HHT (Hilbert–Huang transform) is a method for adaptive analysis of non-linear and non-stationary signals, but it can only distinguish conspicuous faults. SOM (self-organizing feature map) neural network is a network learning with no instructors which has self-adaptive and self-learning features and can compensate for the disadvantage of HHT. This paper proposed a new gear fault identification method based on HHT and SOM neural network. Firstly, the frequency families of gear vibration signals were separated effectively by EMD (empirical mode decomposition). Then Hilbert spectrum and Hilbert marginal spectrum were obtained by Hilbert transform of IMFs (intrinsic mode functions). The amplitude changes of gear vibration signals along with time and frequency had been displayed respectively. After HHT, the energy percentage of the first six IMFs were chosen as input vectors of SOM neural network for fault classification. The analysis results showed that the fault features of these signals can be accurately extracted and distinguished with the proposed approach.     

SOMEDGRA: A case retrieval method for machine tool product configuration design

Case based design is an intelligent method which involves retrieving the most similar previous case to provide a solution of a new decision problem. However, conventional case based design approaches are too reliant on experts’ experiences. A new case retrieval method SOMEDGRA that combines Self-organizing map (SOM) and Euclidean distance (ED) method as well as Grey relational analysis (GRA) method is proposed in case based design. SOM is used to reduce the retrieval range and increase the retrieval efficiency, and ED is used to evaluate the similarity of cases comprehensively. To ensure that the final case has the best overall performance, an evaluation method of similar cases based on GRA is proposed to evaluate similar cases to select the most suitable case. The case study and result on an HTC series machine tool product show that the proposed method is effective, accurate and rapid in the process of product configuration.

     

地铁车辆电动塞拉门的剩余寿命预测

针对地铁车辆客室电动塞拉门传动装置润滑不良的问题,提出了基于自组织映射（SOM）神经网络、隐马尔可夫链（HMC）模型和蒙特卡罗（MC）仿真的剩余使用寿命预测方法。该方法首先对采集到的电机电流信号进行特征提取;然后利用SOM对提取出的多维特征数据进行融合与编码,将所得结果作为HMC的输入向量,训练得到全部寿命下劣化状态转移矩阵;最后利用MC方法实现对其劣化过程的剩余使用寿命预测。故障模拟实验结果表明,该方法可以在考虑润滑不良故障模式下,有效预测得到电动塞拉门丝杆的剩余使用寿命。     

Study of TSP based on self-organizing map

Self-organizing map （SOM） proposed by Kohonen has obtained certain achievements in solving the traveling salesman problem （TSP）.To improve Kohonen SOM,an effective initialization and parameter modification method is discussed to obtain a faster convergence rate and better solution.Therefore,a new improved self-organizing map （ISOM） algorithm is introduced and applied to four traveling salesman problem instances for experimental simulation,and then the result of ISOM is compared with those of four SOM algorithms：AVL,KL,KG and MSTSP.Using ISOM,the average error of four traveling salesman problem instances is only 2.895 0％,which is greatly better than the other four algorithms：8.51％ （AVL）,6.147 5 ％ （KL）,6.555 ％ （KG） and 3.420 9 ％ （MSTSP）.Finally,ISOM is applied to two practical problems：the Chinese 100 cities-TSP and 102 counties-TSP in Shanxi Province,and the two optimal touring routes are provided to the tourists.     

WEAR MONITORING IN TURNING OPERATIONS USING VIBRATION AND STRAIN MEASUREMENTS

For the efficient and reliable operation of automated machining processes, the implementation of suitable tool condition monitoring (TCM) strategy is required. Various monitoring systems, utilising sophisticated signal processing techniques, have been widely researched for a number of different processes. Most monitoring systems developed up to date employ force, acoustic emission and vibration, or a combination of these and other techniques with a sensor integration strategy. With this work, the implementation of a monitoring system utilising simultaneous vibration and strain measurements on the tool tip, is investigated for the wear of synthetic diamond tools which are specifically used for the manufacturing of aluminium pistons. Contrary to many of the earlier investigations, this work was conducted in a manufacturing environment, with the associated constraints such as the impracticality of direct measurement of the wear. Data from the manufacturing process was recorded with two piezoelectric strain sensors and an accelerometer, each coupled to a DSPT Siglab analyser. A large number of features indicative of tool wear were automatically extracted from different parts of the original signals. These included features from the time and frequency domains, time-series model coefficients (as features) and features extracted from wavelet packet analysis. A correlation coefficient approach was used to automatically select the best features indicative of the progressive wear of the diamond tools. The self-organising map (SOM) was employed to identify the tool state. The SOM is a type of neural network based on unsupervised learning. A near 100% correct classification of the tool wear data was obtained by training the SOM with two independent data sets, and testing it with a third independent data set.     

Quantitative identification of illicit drugs by using SOM neural networks

Qualitative identification of THz spectra of illicit drugs using self-organization feature map (SOM) artificial neural network has been demonstrated. In this paper, investigation results show that SOM has quantitatively identified drug mixtures successfully. Based on Beer’s law THz spectra data of various drug proportions were made for training dates. After analyzing the clustering algorithm of SOM, we introduced a parameter named shortest distance as a quantitative criterion for identification result. By this parameter, an automatic recognition algorithm has been developed and successfully applied to the content identification of experimental samples. Combined with our previous work, the SOM neural network can be an integrated and effective method in the identification the THz spectra of illicit drugs.     

离散隐马尔可夫模型在颤振预报中的应用研究

对于切削过程中颤振孕育的动态模式,提出了基于离散隐马尔可夫模型(DHMM)的模式识别理论预报颤振的新方法。首先对切削过程的振动信号进行FFT特征提取,然后利用自组织特征映射(SOM)神经网络对提取的特征矢量进行冗余信息压缩与预分类编码;再根据多变量DHMM建模理论,对切削颤振孕育的各种过程模式建立相应的DHMM,把矢量编码作为观测序列引入到DHMM中进行机器学习、训练;最后将观测序列引入到DHMM中进行颤振孕育的概率识别尝试。实验表明,该方法对颤振孕育过程识别是十分有效的,颤振预报正确率达93.3%。     

基于SOM网络可视化技术的液压系统故障分类研究

针对SOM网络（自组织特征映射神经网络）可视化方法简单、直观的特点,文中将其应用到液压系统的故障分类中。以电流信号的频域能量作为特征参数,用db2共轭正交滤波器组对所获数据进行小波包分解,提取系统在正常及故障运行状态下的特征向量,作为训练样本,然后利用U矩阵图和D矩阵图等可视化T具对分类结果进行仿真与分析,并与一般结果进行比较。结论表明,该方法可行且对故障的判别率高。     

应用SLLE实现手写体数字识别

针对在手写字符识别中由于书写习惯和风格的不同而造成的字符模式不稳定问题,提出了一种基于流形学习的手写体数字识别方法.在流形学习非监督的基础上引入了监督信息,从而保证高维到低维的映射在保留流形某些结构的同时也可进一步分离不同类别的流形.算法首先利用基于监督的局部线性嵌入(SLLE)对手写体数字图像进行字符特征的降维,然后再对降维后的特征进行分类识别.对MIN库中手写体数字数据库进行了实验,实验结果表明,利用SLLE降维以后的特征能够有效地区分字符,识别率可达到93.27%;由于具有较好的识别率,能够发现高维空间的低维嵌入流形.     

Evaluation-based closed-loop errors using principal component analysis and self-organisation map with an application to a pickling process

Closed-loop control is a basic technology in control engineering. Its role is to avoid the tracking error between set points and real variables. The evaluation of plant performance can be based on multivariate statistical process control connected to closed-loop errors behaviour. Due to its practicality, this approach has found many applications in several industries. This paper suggests a combined use of principal component analysis (PCA) and self-organisation map (SOM) algorithms to evaluate the process on the basis of closed-loop errors dynamic. Generally, it is possible to evaluate a product quality in the basis of the dynamic changes of the closed-loop control errors. In this paper, a new method based on the analysis of the control errors is proposed; it is carried out by a combined use of the PCA-SOM algorithm. Comparatively to the conventional PCA method, this new technique is characterised by the performant indexes that give an accurate evaluation of the process variability and its impact on the product quality. As shown in the different simulation results, the proposed approach gives a global evaluation and improves considerably the performance of computed indexes used for the evaluation of the controlled process.     

SOM-BP神经网络在水稻钵育栽植机关键部件设计中的应用初探

水稻钵育栽植机的设计是实现水稻钵育栽培技术的关键环节,区别于传统的水稻栽植机其关键部件的设计方法还不成熟,在理论研究方面还不完善。为探索新的设计方法,论述了神经网络算法对农业机械设计的适用性,以自组织特征映射网络SOM和误差反向传播网络BP为理论基础,将SOM-BP集成神经网络模型应用于水稻钵育栽植机关键部件设计领域,建立神经网络模型,通过集成网络训练,得到设计结果。验证了SOM-BP神经网络在农业机械机设计中的正确性和精确性。     

基于KPCA-SOM网络的列控RBC系统故障诊断方法

无线闭塞中心(RBC)系统是CTCS-3级列控系统的核心设备,在现场其故障分析主要依靠人工完成,诊断结果不精确、效率低。因此,提出了基于one-hot模型、核主元分析(KPCA)和自组织映射(SOM)网络的RBC系统智能故障诊断方法。首先,通过人工选取的故障特征词库和故障追踪记录表构建基于“one-hot”模型的故障文档矩阵;其次,利用核主元分析方法对故障文档矩阵进行降维降噪处理,避免信息冗余;最后将处理后的数据输入至SOM网络,训练生成KPCA-SOM故障分类模型。通过与BP神经网络算法、SOM网络算法比对分析,KPCA-SOM智能诊断方法可有效地对列控RBC系统常见故障类型进行区分,并且在准确率和处理效率上进一步优化提升。     

ON LINE MONITORING OF BURNING THROUGH FOR SHORT CIRCUIT CO_2 ARC WELDING BASED ON THE SELF-ORGANIZE FEATURE MAP NEURAL NETWORKS

0　ＩＮＴＲＯＤＵＣＴＩＯＮＡｍｏｎｇｔｈｅｖａｒｉｅｔｙｏｆｆｕｓｉｏｎｗｅｌｄｉｎｇｐｒｏｃｅｓｓｅｓａｖａｉｌａｂｌｅ,ｓｈｏｒｔｃｉｒｃｕｉｔＣＯ2 ａｒｃｗｅｌｄｉｎｇｉｓａｔｅｃｈｎｏｌｏｇｙｗｈｉｃｈｉｓｏｎｅｏｆｔｈｅｍｏｓｔｆｒｅｑｕｅｎｔｌｙｕｓｅｄｍｅｔｈｏｄｉｎａｗｉｄｅｒａｎｇｅｏｆａｐｐｌｉｃａｔｉｏｎｓａｎｄａｌｓｏａｕｔｏｍａｔｉｏｎｂｅｃａｕｓｅｏｆｉｔｓｖｅｒｓａｔｉｌｉｔｙａｎｄｃｏｓｔｅｆｆｅｃｔｉｖｅｎｅｓｓ.Ｔｈｅｒｅｆｏｒｅ ,ｓｔｕｄｙｏｎｔｈｅｐｒｉｎ…     

FAULT DIAGNOSIS BASED ON INTEGRATION OF CLUSTER ANALYSIS, ROUGH SET METHOD AND FUZZY NEURAL NETWORK

In order to increase the efficiency and decrease the cost of machinery diagnosis, a hybrid system of computational intelligence methods is presented. Firstly, the continuous attributes in diagnosis decision system are discretized with the self-organizing map (SOM) neural network. Then, dynamic reducts are computed based on rough set method, and the key conditions for diagnosis are found according to the maximum cluster ratio. Lastly, according to the optimal reduct, the adaptive neuro-fuzzy inference system (ANFIS) is designed for fault identification. The diagnosis of a diesel verifies the feasibility of engineering applications.     

基于SOM神经网络的摩擦状态识别研究

为提取摩擦振动的特征和实现摩擦副摩擦状态的识别,在往复摩擦磨损试验机进行摩擦副混合摩擦和干摩擦状态的摩擦磨损试验。应用谱减法对试验采集的摩擦振动信号进行降噪,计算降噪后的摩擦振动15个特征参数。应用自组织映射(Self-organizing map, SOM)神经网络对摩擦副不同摩擦状态的摩擦振动特征参数进行分析,得到摩擦振动的SOM神经网络神经元分类。研究结果表明,谱减法能消除摩擦磨损试验机的背景噪声,SOM神经网络算法能够有效分析摩擦振动信号的特征,实现摩擦副摩擦状态的识别。     

基于SOM网络的特征选择技术研究

讨论了一种SOM网络训练结果的可视化技巧,结合该技巧提出了基于SOM网络的特征选择方法。该方法 通过计算出SOM网络竞争层神经元权值中各维特征对输入模式聚类识别的影响,可以选择出对于模式识别敏感 的特征集。用IRIS和齿轮故障数据对该方法进行了检验,研究结果表明,采用该方法能较好地从原始特征中选择 出有效特征子集,实现不同类别输入数据之间的模式聚类识别。