基于多尺度熵和距离评估的滚动轴承故障诊断

为了解决滚动轴承非线性故障特征难以提取的问题,将多尺度熵和距离评估方法有机结合,提出一种新的滚动轴承非线性特征提取方法。首先利用多尺度熵来刻画滚动轴承振动信号在不同尺度上的细节特征;然后引入距离评估选择算法,对多尺度特征进行特征评估,优化选取敏感尺度特征,并送入支持向量机实现轴承不同状态的分类。通过滚动轴承故障诊断实验对该方法进行验证,结果表明,选取敏感多尺度熵特征集对故障特征进行刻画,较单尺度特征、原始特征集和随机特征集等诊断效果更佳,提高了诊断精度。     

Design and comparison of some Shewhart‐type schemes for simultaneous monitoring of Weibull parameters

Weibull distribution is one of the most important probability models used in modeling time between events, system reliability, and particle sizes, among others. Therefore, efficiently and consequently monitoring certain changes in Weibull process is considered as an important research topic. Various statistical process monitoring schemes have been developed for monitoring different process parameters, including some for Weibull parameters. Most of these schemes are, however, designed to monitor and control a single process parameter, although there are two important model parameters for Weibull distribution. Recently, several researchers studied various schemes for jointly monitoring the mean and variance of a normally distributed process using a single plotting statistic. Nevertheless, there is still dearth of researches in joint monitoring of non‐normal process parameters. In this context, we develop some control schemes for simultaneously monitoring the scale and shape parameters of processes that follow the Weibull distribution. Implementation procedures are developed, and performance properties of various proposed schemes are investigated. We also offer an illustrative example along with a summary and recommendations.     

Designing a multivariate–multistage quality control system using artificial neural networks

In most real-world manufacturing systems, the production of goods comprises several autocorrelated stages and the quality characteristics of the goods at each stage are correlated random variables. This paper addresses the problem of monitoring a multivariate–multistage manufacturing process and diagnoses the possible causes of out-of-control signals. To achieve this purpose using multivariate time series models, first a model for the autocorrelated data coming from multivariate–multistage processes is developed. Then, a single neural network is designed, trained and employed to control and classify mean shifts in quality characteristics of all stages. In-control and out-of-control average run lengths and correct classification ratio indices have been chosen to investigate the performance of the designed network. The results of a simulation study show that the network is capable of detecting both in-control and out-of-control signals appropriately.     

Diagnosing the variance shifts signal in multivariate process control using ensemble classifiers

Statistical process control charts have been successfully used to monitor process stability in various industries. The need to simultaneously monitor two or more quality characteristics has led to the prevalent adoption of multivariate control charts. However, out-of-control signals in multivariate control charts may be caused by one or more variables, or a set of variables. Therefore, effective quality control requires not only the rapid detection of process fluctuations, but also the correct identification of the variable(s) responsible for those changes. This study approaches the diagnosis of out-of-control signals as a classification task and proposes a support vector machine (SVM)-based ensemble classification model focused on variance shifts in multivariate processes. We address the issues of data diversity and ensemble method in constructing an ensemble model. Simulation results demonstrate the effectiveness of the proposed ensemble classification model in identifying the source of variance change. The proposed method clearly outperforms single classifiers as well as other comparable models including bagging and boosting. The results also reveal that the use of extracted features as input vectors for SVM provides better classification performance than the use of raw data. The proposed SVM-based ensemble classification system provides a reliable tool for the interpretation of out-of-control signals in multivariate process control.     

A comparison study of control charts for statistical monitoring of functional data

The quality of products and processes is more and more often becoming related to functional data, which refer to information summarised in the form of profiles. The recent literature has pointed out that traditional control charting methods cannot be directly applied in these cases and new approaches for profile monitoring are required. While many different profile monitoring approaches have been proposed in the scientific literature, few comparison studies are available. This paper aims at filling this gap by comparing three representative profile monitoring approaches in different production scenarios. The performance comparison will allow us to select a specific approach in a given situation. The competitor approaches are chosen to represent different levels of complexity, as well as different types of modelling approaches. In particular, at a lower level of complexity, the ‘location control chart’ (where the upper and lower control limits are ±K standard deviations from the sample mean at each profile location) is considered to be representative of industrial practice. At a higher complexity level, approaches based on combining a parametric model of functional data with multivariate and univariate control charting are considered. Within this second class, we analyse two different approaches. The first is based on regression and the second focuses on using principal component analysis for modelling functional data. A manufacturing reference case study is used throughout the paper, namely profiles measured on machined items subject to geometrical specification (roundness).     

A hybrid learning-based model for on-line monitoring and diagnosis of out-of-control signals in multivariate manufacturing processes

In multivariate statistical process control, most multivariate quality control charts are shown to be effective in detecting out-of-control signals based upon overall statistics. But these charts do not relieve the need for pinpointing the source(s) of the out-of-control signals. In addition, these charts cannot provide more detailed process information, such as quantitative abnormal assessment values and visualisation of process changes, which would be very useful for quality practitioners to locate the assignable causes that give rise to the out-of-control situation. In this study, a hybrid learning-based model has been investigated for monitoring and diagnosing out-of-control signals in a bivariate process. In this model, a minimum quantisation error (MQE) chart based on the self-organization map (SOM) neural network (NN) was developed for monitoring process changes (i.e., mean shifts), and a selective NN ensemble approach (DPSOEN) was developed for diagnosing signals that are judged as out-of-control signals by MQE charts. The simulation results demonstrate that the proposed model outperforms the conventional multivariate control scheme in terms of average run length (ARL), and can accurately classify the source(s) of out-of-control signals. An extensive experiment is also carried out to examine the effects of six statistical features on the performance of DPSOEN.     

Comprehensive control charting applied to chromatography

Multivariate statistical process control (MSPC) based for example on principal component analysis (PCA) can make use of the information contained in multiple measured signals simultaneously. This can be much more powerful in detecting variations due to special causes than conventional single variable statistical process control (SPC). Furthermore, the PCA based SPC simplifies monitoring as it limits the number of control charts to typically two charts rather than one for each signal. However, the derived MSPC statistics may suffer from lack of sensitivity if only one or a few variables deviate in a given situation. In this paper we develop a new comprehensive control (COCO) chart procedure that considers both univariate statistics and multivariate statistics derived from PCA in a single plot that allows easy visualization of the combined data from a univariate and multivariate point of view. The method is exemplified using twenty analytical chromatographic peak areas obtained for purity analysis of a biopharmaceutical drug substance. The new control chart procedure detected two different types of faulty events in this study.     

A class of new nonparametric circular-grid charts for signal classification

In this paper, some efficient monitoring and post-signal follow-up approaches are studied and compared for joint surveillance of location and scale of a process using the notions of circular-grid (CG) schemes. Precisely, three variants of CG Cucconi schemes are introduced and compared with three variants of percentile modified Lepage (PML) schemes. One of the PML schemes is equivalent to the traditional CG Lepage scheme, while another may be viewed as the Lepage type statistic using Gastwrith score, which is also a powerful tool for process surveillance. Overall, one of the proposed CG Cucconi schemes is most effective in identifying a class of signals, whether it is a location shift or scale shift or a shift in both parameters. It also indicates the direction of the shifts in either or both the parameters. Detecting a downward scale shift is the most challenging task in joint monitoring, and to this end, a new bias-corrected CG Lepage scheme is introduced. We compare the competing schemes in terms of correct signal classification probabilities. We illustrate the use of the proposed schemes in monitoring the trip-duration data in cab services. Some concluding remarks and future research problems are offered.     

A unified statistical framework for monitoring multivariate systems with unknown source and error signals

This article proposes a unified multivariate statistical monitoring framework that incorporates recent work on maximum likelihood PCA (MLPCA) into conventional PCA-based monitoring. The proposed approach allows the simultaneous and consistent estimation of the PCA model plane, its dimension and the error covariance matrix. This paper also invokes recent work on monitoring non-Gaussian processes to extract unknown Gaussian as well as non-Gaussian source signals from recorded process data. By contrasting the unified framework with PCA-based process monitoring using a simulation example and recorded data from two industrial processes, the proposed approach produced more accurate and/or sensitive monitoring models.     

Process analysis,monitoring and diagnosis,using multivariate projection methods

Multivariate statistical methods for the analysis, monitoring and diagnosis of process operating performance are becoming more important because of the availability of on-line process computers which routinely collect measurements on large numbers of process variables. Traditional univariate control charts have been extended to multivariate quality control situations using the Hotelling T² statistic. Recent approaches to multivariate statistical process control which utilize not only product quality data (Y), but also all of the available process variable data (X) are based on multivariate statistical projection methods (principal component analysis, (PCA), partial least squares, (PLS), multi-block PLS and multi-way PCA). An overview of these methods and their use in the statistical process control of multivariate continuous and batch processes is presented. Applications are provided on the analysis of historical data from the catalytic cracking section of a large petroleum refinery, on the monitoring and diagnosis of a continuous polymerization process and on the monitoring of an industrial batch process.     

Variable Selection‐based Multivariate Cumulative Sum Control Chart

High‐dimensional applications pose a significant challenge to the capability of conventional statistical process control techniques in detecting abnormal changes in process parameters. These techniques fail to recognize out‐of‐control signals and locate the root causes of faults especially when small shifts occur in high‐dimensional variables under the sparsity assumption of process mean changes. In this paper, we propose a variable selection‐based multivariate cumulative sum (VS‐MCUSUM) chart for enhancing sensitivity to out‐of‐control conditions in high‐dimensional processes. While other existing charts with variable selection techniques tend to show weak performances in detecting small shifts in process parameters due to the misidentification of the ‘faulty’ parameters, the proposed chart performs well for small process shifts in identifying the parameters. The performance of the VS‐MCUSUM chart under different combinations of design parameters is compared with the conventional MCUSUM and the VS‐multivariate exponentially weighted moving average control charts. Finally, a case study is presented as a real‐life example to illustrate the operational procedures of the proposed chart. Both the simulation and numerical studies show the superior performance of the proposed chart in detecting mean shift in multivariate processes. Copyright © 2016 John Wiley & Sons, Ltd.     

Multivariate image analysis: A review with applications

Nowadays, image analysis is becoming more important because of its ability to perform fast and non-invasive low-cost analysis on products and processes. Image analysis is a wide denomination that encloses classical studies on gray scale or RGB images, analysis of images collected using few spectral channels (sometimes called multispectral images) or, most recently, data treatments to deal with hyperspectral images, where the spectral direction is exploited in its full extension. Pioneering data treatments in image analysis were applied to simple images mainly for defect detection, segmentation and classification by the Computer Science community. From the late 80s, the chemometric community joined this field introducing powerful tools for image analysis, which were already in use for the study of classical spectroscopic data sets and were appropriately modified to fit the particular characteristics of image structures. These chemometric approaches adapt to images of all kinds, from the simplest to the hyperspectral images, and have provided new insights on the spatial and spectroscopic information of this kind of data sets. New fields open by the introduction of chemometrics on image analysis are exploratory image analysis, multivariate statistical process control (monitoring), multivariate image regression or image resolution. This paper reviews the different techniques developed in image analysis and shows the evolution in the information provided by the different methodologies, which has been heavily pushed by the increasing complexity of the image measurements in the spatial and, particularly, in the spectral direction.     

Fault detection based on Gaussian process latent variable models

Gaussian processes, GPs, can be used to approximate complex non-linear functions with relative simplicity. Their regression performance is, at least, comparable to that achieved via artificial neural networks (ANN) and, in fact, both methods are intrinsically related. They are both non-parametric and, as Neal (1994) [1] has shown, when the number of nodes in the hidden layer of a neural network tends to infinity the ANN converge to a Gaussian process.In most of the cases, the GP will map a multivariate input into a univariate response. In this paper, however, we present an approach to process monitoring that combines several GPs so that multivariate responses can be appropriately modeled. We review a similar approach recently proposed in the literature and highlight some concerns related to it that needs to be taken into consideration. Additionally, we propose an alternative procedure to the way in which new observations are mapped into the non-linear model. A simulation study is provided that will help understand the method flexibility. Furthermore, results from a real example are also discussed.     

A New Multivariate Process Capability Index Under Both Unilateral and Bilateral Quality Characteristics

Process capability indices (PCIs) are used in statistical process control to evaluate the capability of the processes in satisfying the customer's needs. In the past two decades varieties of PCI are introduced by researchers to analyze the process capability with univariate or multivariate quality characteristics. To the best of our knowledge, most famous multivariate capability indices are proposed when the quality characteristics have both upper and lower specification limits. These indices are incapable to assess the multivariate processes capability with unilateral specification. In this article, we propose a new multivariate PCI to analyze the processes with one or more unilateral specification limits. This new index also accounts for all problems in the best PCIs of the literature. The performance of the proposed index is evaluated by real cases under different situations. The results show that the proposed index performs satisfactorily in all cases considered. Copyright © 2012 John Wiley & Sons, Ltd.     

In situ process monitoring of hierarchical micro-/nano-composites using percolated carbon nanotube networks

We report a method to monitor the manufacturing process of hierarchical micro-/nano-composites that uses integrated and percolated multi-walled carbon nanotube (MWCNT) networks with the aim of reducing part-to-part variability. Composites were fabricated by VARTM. Fiber textiles were spray-coated with MWCNTs, electrodes were embedded prior to vacuum bagging. In situ process monitoring was achieved by measurement of the electrical resistance of electrode pairs. The effects of MWCNT density and length on the ability to monitor the manufacturing process were evaluated. Experiments showed that monitoring the changes in resistance between electrode pairs on the conductive MWCNT network allowed various events during the manufacturing process, including part infusion, onset of crosslinking, and gel point of the resin, which are necessary for accurate evaluation of part quality. Our simple yet effective method to monitor the manufacturing processes and predict the final-part quality of multiscale composites can be integrated into existing processes with minimal modifications.     

A Hybrid Learning‐Based Model for Simultaneous Monitoring of Process Mean and Variance

Unnatural patterns exhibited on process mean and variance control charts can be associated separately with different assignable causes. Quick and accurate knowledge of the type of control chart patterns (CCPs), either because of process mean or variance, can greatly facilitate identification of assignable causes. Over the past few decades, however, process mean and variance CCPs are seldom studied simultaneously in the statistical process control literature. This study proposes a hybrid learning‐based model for simultaneous monitoring of process mean and variance CCPs. In this model, a self‐organization map neural network‐based quantization error control chart is responsible for detecting the out‐of‐control signals, a discrete particle swarm optimization‐based selective ensemble of back‐propagation networks is responsible for classifying the detected out‐of‐control signals into categories of mean and/or variance abnormality, and two discrete particle swarm optimization‐based selective ensembles of learning vector quantization networks are responsible for further identifying the detected mean and variance out‐of‐control signals as one of the specific CCP types, respectively. Extensive simulations indicate that the proposed hybrid learning‐based model outperforms other existing approaches in detecting mean and variance changes, while also capable of CCP recognition. In addition, a case study is conducted to demonstrate how the proposed hybrid learning‐based model can function as an effective tool for monitoring mean and variance simultaneously. Copyright © 2013 John Wiley & Sons, Ltd.     

A variable‐selection control chart via penalized likelihood and Gaussian mixture model for multimodal and high‐dimensional processes

With the development of the sensor network and manufacturing technology, multivariate processes face a new challenge of high‐dimensional data. However, traditional statistical methods based on small‐ or medium‐sized samples such as T² monitoring statistics may not be suitable because of the “curse of dimensionality” problem. To overcome this shortcoming, some control charts based on the variable‐selection (VS) algorithms using penalized likelihood have been suggested for process monitoring and fault diagnosis. Although there has been much effort to improve VS‐based control charts, there is usually a common distributional assumption that in‐control observations should follow a single multivariate Gaussian distribution. However, in current manufacturing processes, processes can have multimodal properties. To handle the high‐dimensionality and multimodality, in this study, a VS‐based control chart with a Gaussian mixture model (GMM) is proposed. We extend the VS‐based control chart framework to the process with multimodal distributions, so that the high‐dimensionality and multimodal information in the process can be better considered.     

Distribution‐free Lepage Type Circular‐grid Charts for Joint Monitoring of Location and Scale Parameters of a Process

In the last 5 years, research works on distribution‐free (nonparametric) process monitoring have registered a phenomenal growth. A Google Scholar database search on early September 2015 reveals 246 articles on distribution‐free control charts during 2000–2009 and 466 articles in the following years. These figures are about 1400 and 2860 respectively if the word ‘nonparametric’ is used in place of ‘distribution‐free’. Distribution‐free charts do not require any prior knowledge about the process parameters. Consequently, they are very effective in monitoring various non‐normal and complex processes. Traditional process monitoring schemes use two separate charts, one for monitoring process location and the other for process scale. Recently, various schemes have been introduced to monitor the process location and process scale simultaneously using a single chart. Performance advantages of such charts have been clearly established. In this paper, we introduce a new graphical device, namely, circular‐grid charts, for simultaneous monitoring of process location and process scale based on Lepage‐type statistics. We also discuss general form of Lepage statistics and show that a new modified Lepage statistic is often better than the traditional of Lepage statistic. We offer a new and attractive post‐signal follow‐up analysis. A detailed numerical study based on Monte‐Carlo simulations is performed, and some illustrations are provided. A clear guideline for practitioners is offered to facilitate the best selection of charts among various alternatives for simultaneous monitoring of location‐scale. The practical application of the charts is illustrated. Copyright © 2016 John Wiley & Sons, Ltd.